id sid tid token lemma pos fis-4216 1 1 shot shoot VERB fis-4216 1 2 peening peene VERB fis-4216 1 3 processes process NOUN fis-4216 1 4 to to PART fis-4216 1 5 obtain obtain VERB fis-4216 1 6 nanocrystalline nanocrystalline ADJ fis-4216 1 7 surfaces surface NOUN fis-4216 1 8 in in ADP fis-4216 1 9 metal metal NOUN fis-4216 1 10 alloys alloy NOUN fis-4216 1 11 : : PUNCT fis-4216 1 12 l. l. PROPN fis-4216 1 13 wang wang PROPN fis-4216 1 14 , , PUNCT fis-4216 1 15 frattura frattura PROPN fis-4216 1 16 ed ed PROPN fis-4216 1 17 integrità integrità PROPN fis-4216 1 18 strutturale strutturale PROPN fis-4216 1 19 , , PUNCT fis-4216 1 20 65 65 NUM fis-4216 1 21 ( ( PUNCT fis-4216 1 22 2023 2023 NUM fis-4216 1 23 ) ) PUNCT fis-4216 1 24 289 289 NUM fis-4216 1 25 - - SYM fis-4216 1 26 299 299 NUM fis-4216 1 27 ; ; PUNCT fis-4216 1 28 doi doi NOUN fis-4216 1 29 : : PUNCT fis-4216 1 30 10.3221 10.3221 NUM fis-4216 1 31 / / SYM fis-4216 1 32 igf igf PROPN fis-4216 1 33 - - PUNCT fis-4216 1 34 esis.65.19 esis.65.19 NOUN fis-4216 1 35 289 289 NUM fis-4216 1 36 automatic automatic ADJ fis-4216 1 37 detection detection NOUN fis-4216 1 38 of of ADP fis-4216 1 39 concrete concrete ADJ fis-4216 1 40 cracks crack NOUN fis-4216 1 41 from from ADP fis-4216 1 42 images image NOUN fis-4216 1 43 using use VERB fis-4216 1 44 adamsqueezenet adamsqueezenet NOUN fis-4216 1 45 deep deep ADJ fis-4216 1 46 learning learning NOUN fis-4216 1 47 model model NOUN fis-4216 1 48 lin lin PROPN fis-4216 1 49 wang wang PROPN fis-4216 1 50 hefei hefei PROPN fis-4216 1 51 university university PROPN fis-4216 1 52 , , PUNCT fis-4216 1 53 hefei hefei PROPN fis-4216 1 54 , , PUNCT fis-4216 1 55 anhui anhui PROPN fis-4216 1 56 province province PROPN fis-4216 1 57 , , PUNCT fis-4216 1 58 230061 230061 NUM fis-4216 1 59 , , PUNCT fis-4216 1 60 china china PROPN fis-4216 1 61 linw@smu.edu linw@smu.edu PROPN fis-4216 1 62 , , PUNCT fis-4216 1 63 http://orcid.org/0000-0002-6481-2409 http://orcid.org/0000-0002-6481-2409 PROPN fis-4216 1 64 abstract abstract NOUN fis-4216 1 65 . . PUNCT fis-4216 2 1 cracks crack NOUN fis-4216 2 2 in in ADP fis-4216 2 3 the the DET fis-4216 2 4 concrete concrete ADJ fis-4216 2 5 surfaces surface NOUN fis-4216 2 6 are be AUX fis-4216 2 7 typically typically ADV fis-4216 2 8 clear clear ADJ fis-4216 2 9 warning warning NOUN fis-4216 2 10 signs sign NOUN fis-4216 2 11 of of ADP fis-4216 2 12 a a DET fis-4216 2 13 potential potential ADJ fis-4216 2 14 threat threat NOUN fis-4216 2 15 to to ADP fis-4216 2 16 the the DET fis-4216 2 17 integrity integrity NOUN fis-4216 2 18 and and CCONJ fis-4216 2 19 serviceability serviceability NOUN fis-4216 2 20 of of ADP fis-4216 2 21 the the DET fis-4216 2 22 structure structure NOUN fis-4216 2 23 . . PUNCT fis-4216 3 1 the the DET fis-4216 3 2 techniques technique NOUN fis-4216 3 3 based base VERB fis-4216 3 4 on on ADP fis-4216 3 5 image image NOUN fis-4216 3 6 processing processing NOUN fis-4216 3 7 can can AUX fis-4216 3 8 effectively effectively ADV fis-4216 3 9 detect detect VERB fis-4216 3 10 cracks crack NOUN fis-4216 3 11 in in ADP fis-4216 3 12 digital digital ADJ fis-4216 3 13 images image NOUN fis-4216 3 14 . . PUNCT fis-4216 4 1 these these DET fis-4216 4 2 techniques technique NOUN fis-4216 4 3 , , PUNCT fis-4216 4 4 however however ADV fis-4216 4 5 , , PUNCT fis-4216 4 6 are be AUX fis-4216 4 7 generally generally ADV fis-4216 4 8 susceptible susceptible ADJ fis-4216 4 9 to to ADP fis-4216 4 10 user user NOUN fis-4216 4 11 - - PUNCT fis-4216 4 12 driven drive VERB fis-4216 4 13 heuristic heuristic ADJ fis-4216 4 14 thresholds threshold NOUN fis-4216 4 15 and and CCONJ fis-4216 4 16 irrelevant irrelevant ADJ fis-4216 4 17 distractors distractor NOUN fis-4216 4 18 . . PUNCT fis-4216 5 1 inspired inspire VERB fis-4216 5 2 by by ADP fis-4216 5 3 the the DET fis-4216 5 4 recent recent ADJ fis-4216 5 5 success success NOUN fis-4216 5 6 of of ADP fis-4216 5 7 artificial artificial ADJ fis-4216 5 8 intelligence intelligence NOUN fis-4216 5 9 , , PUNCT fis-4216 5 10 a a DET fis-4216 5 11 deep deep ADJ fis-4216 5 12 learning learning NOUN fis-4216 5 13 - - PUNCT fis-4216 5 14 based base VERB fis-4216 5 15 automated automate VERB fis-4216 5 16 crack crack NOUN fis-4216 5 17 detection detection NOUN fis-4216 5 18 system system NOUN fis-4216 5 19 named name VERB fis-4216 5 20 cracksn cracksn NOUN fis-4216 5 21 was be AUX fis-4216 5 22 presented present VERB fis-4216 5 23 . . PUNCT fis-4216 6 1 this this DET fis-4216 6 2 proposed propose VERB fis-4216 6 3 deep deep ADJ fis-4216 6 4 learning learning NOUN fis-4216 6 5 model model NOUN fis-4216 6 6 , , PUNCT fis-4216 6 7 built build VERB fis-4216 6 8 on on ADP fis-4216 6 9 the the DET fis-4216 6 10 adam adam NOUN fis-4216 6 11 - - PUNCT fis-4216 6 12 squeezenet squeezenet NOUN fis-4216 6 13 architecture architecture NOUN fis-4216 6 14 , , PUNCT fis-4216 6 15 automatically automatically ADV fis-4216 6 16 learns learn VERB fis-4216 6 17 the the DET fis-4216 6 18 discriminative discriminative NOUN fis-4216 6 19 feature feature NOUN fis-4216 6 20 directly directly ADV fis-4216 6 21 from from ADP fis-4216 6 22 the the DET fis-4216 6 23 labeled label VERB fis-4216 6 24 and and CCONJ fis-4216 6 25 augmented augment VERB fis-4216 6 26 patches patch NOUN fis-4216 6 27 . . PUNCT fis-4216 7 1 an an DET fis-4216 7 2 image image NOUN fis-4216 7 3 dataset dataset NOUN fis-4216 7 4 of of ADP fis-4216 7 5 concrete concrete ADJ fis-4216 7 6 surfaces surface NOUN fis-4216 7 7 is be AUX fis-4216 7 8 collected collect VERB fis-4216 7 9 by by ADP fis-4216 7 10 smartphone smartphone NOUN fis-4216 7 11 and and CCONJ fis-4216 7 12 carefully carefully ADV fis-4216 7 13 prepared prepare VERB fis-4216 7 14 in in ADP fis-4216 7 15 order order NOUN fis-4216 7 16 to to PART fis-4216 7 17 develop develop VERB fis-4216 7 18 and and CCONJ fis-4216 7 19 train train VERB fis-4216 7 20 the the DET fis-4216 7 21 cracksn cracksn NOUN fis-4216 7 22 system system NOUN fis-4216 7 23 . . PUNCT fis-4216 8 1 the the DET fis-4216 8 2 hyperparameters hyperparameter NOUN fis-4216 8 3 of of ADP fis-4216 8 4 squeezenet squeezenet NOUN fis-4216 8 5 are be AUX fis-4216 8 6 tuned tune VERB fis-4216 8 7 with with ADP fis-4216 8 8 adam adam PROPN fis-4216 8 9 optimization optimization NOUN fis-4216 8 10 through through ADP fis-4216 8 11 the the DET fis-4216 8 12 training training NOUN fis-4216 8 13 and and CCONJ fis-4216 8 14 validation validation NOUN fis-4216 8 15 procedures procedure NOUN fis-4216 8 16 . . PUNCT fis-4216 9 1 the the DET fis-4216 9 2 fine fine ADV fis-4216 9 3 - - PUNCT fis-4216 9 4 tuned tune VERB fis-4216 9 5 cracksn cracksn NOUN fis-4216 9 6 system system NOUN fis-4216 9 7 outperforms outperform VERB fis-4216 9 8 state state NOUN fis-4216 9 9 - - PUNCT fis-4216 9 10 of of ADP fis-4216 9 11 - - PUNCT fis-4216 9 12 the the DET fis-4216 9 13 - - PUNCT fis-4216 9 14 art art NOUN fis-4216 9 15 models model NOUN fis-4216 9 16 in in ADP fis-4216 9 17 recent recent ADJ fis-4216 9 18 literature literature NOUN fis-4216 9 19 by by ADP fis-4216 9 20 correctly correctly ADV fis-4216 9 21 classifying classify VERB fis-4216 9 22 97.3 97.3 NUM fis-4216 9 23 % % NOUN fis-4216 9 24 of of ADP fis-4216 9 25 the the DET fis-4216 9 26 cracked crack VERB fis-4216 9 27 patches patch NOUN fis-4216 9 28 in in ADP fis-4216 9 29 the the DET fis-4216 9 30 image image NOUN fis-4216 9 31 dataset dataset NOUN fis-4216 9 32 . . PUNCT fis-4216 10 1 the the DET fis-4216 10 2 success success NOUN fis-4216 10 3 of of ADP fis-4216 10 4 cracksn cracksn NOUN fis-4216 10 5 , , PUNCT fis-4216 10 6 demonstrated demonstrate VERB fis-4216 10 7 by by ADP fis-4216 10 8 its its PRON fis-4216 10 9 light light ADJ fis-4216 10 10 network network NOUN fis-4216 10 11 design design NOUN fis-4216 10 12 and and CCONJ fis-4216 10 13 outstanding outstanding ADJ fis-4216 10 14 performance performance NOUN fis-4216 10 15 , , PUNCT fis-4216 10 16 provides provide VERB fis-4216 10 17 a a DET fis-4216 10 18 crucial crucial ADJ fis-4216 10 19 step step NOUN fis-4216 10 20 toward toward ADP fis-4216 10 21 automated automate VERB fis-4216 10 22 damage damage NOUN fis-4216 10 23 inspection inspection NOUN fis-4216 10 24 and and CCONJ fis-4216 10 25 health health NOUN fis-4216 10 26 evaluation evaluation NOUN fis-4216 10 27 for for ADP fis-4216 10 28 infrastructure infrastructure NOUN fis-4216 10 29 . . PUNCT fis-4216 11 1 keywords keyword NOUN fis-4216 11 2 . . PUNCT fis-4216 12 1 concrete concrete ADJ fis-4216 12 2 crack crack NOUN fis-4216 12 3 , , PUNCT fis-4216 12 4 automated automate VERB fis-4216 12 5 damage damage NOUN fis-4216 12 6 inspection inspection NOUN fis-4216 12 7 , , PUNCT fis-4216 12 8 squeezenet squeezenet NOUN fis-4216 12 9 , , PUNCT fis-4216 12 10 adam adam PROPN fis-4216 12 11 optimization optimization PROPN fis-4216 12 12 , , PUNCT fis-4216 12 13 deep deep ADJ fis-4216 12 14 learning learning NOUN fis-4216 12 15 . . PUNCT fis-4216 13 1 citation citation NOUN fis-4216 13 2 : : PUNCT fis-4216 14 1 wang wang PROPN fis-4216 14 2 , , PUNCT fis-4216 14 3 l. l. PROPN fis-4216 14 4 , , PUNCT fis-4216 14 5 automatic automatic ADJ fis-4216 14 6 detection detection NOUN fis-4216 14 7 of of ADP fis-4216 14 8 concrete concrete ADJ fis-4216 14 9 crack crack NOUN fis-4216 14 10 from from ADP fis-4216 14 11 images image NOUN fis-4216 14 12 using use VERB fis-4216 14 13 adamsqueezenet adamsqueezenet NOUN fis-4216 14 14 deep deep ADJ fis-4216 14 15 learning learning NOUN fis-4216 14 16 model model NOUN fis-4216 14 17 , , PUNCT fis-4216 14 18 frattura frattura PROPN fis-4216 14 19 ed ed PROPN fis-4216 14 20 integrità integrità PROPN fis-4216 14 21 strutturale strutturale PROPN fis-4216 14 22 , , PUNCT fis-4216 14 23 65 65 NUM fis-4216 14 24 ( ( PUNCT fis-4216 14 25 2023 2023 NUM fis-4216 14 26 ) ) PUNCT fis-4216 14 27 289 289 NUM fis-4216 14 28 - - SYM fis-4216 14 29 299 299 NUM fis-4216 14 30 . . PUNCT fis-4216 14 31 received receive VERB fis-4216 14 32 : : PUNCT fis-4216 14 33 07.04.2023 07.04.2023 NUM fis-4216 14 34 accepted accept VERB fis-4216 14 35 : : PUNCT fis-4216 14 36 18.06.2023 18.06.2023 NUM fis-4216 14 37 online online ADJ fis-4216 14 38 first first ADV fis-4216 14 39 : : PUNCT fis-4216 14 40 19.06.2023 19.06.2023 NUM fis-4216 14 41 published publish VERB fis-4216 14 42 : : PUNCT fis-4216 14 43 01.07.2023 01.07.2023 NUM fis-4216 14 44 copyright copyright NOUN fis-4216 14 45 : : PUNCT fis-4216 15 1 © © PROPN fis-4216 15 2 2023 2023 NUM fis-4216 15 3 this this PRON fis-4216 15 4 is be AUX fis-4216 15 5 an an DET fis-4216 15 6 open open ADJ fis-4216 15 7 access access NOUN fis-4216 15 8 article article NOUN fis-4216 15 9 under under ADP fis-4216 15 10 the the DET fis-4216 15 11 terms term NOUN fis-4216 15 12 of of ADP fis-4216 15 13 the the DET fis-4216 15 14 cc cc NOUN fis-4216 15 15 - - PUNCT fis-4216 15 16 by by ADP fis-4216 15 17 4.0 4.0 NUM fis-4216 15 18 , , PUNCT fis-4216 15 19 which which PRON fis-4216 15 20 permits permit VERB fis-4216 15 21 unrestricted unrestricted ADJ fis-4216 15 22 use use NOUN fis-4216 15 23 , , PUNCT fis-4216 15 24 distribution distribution NOUN fis-4216 15 25 , , PUNCT fis-4216 15 26 and and CCONJ fis-4216 15 27 reproduction reproduction NOUN fis-4216 15 28 in in ADP fis-4216 15 29 any any DET fis-4216 15 30 medium medium NOUN fis-4216 15 31 , , PUNCT fis-4216 15 32 provided provide VERB fis-4216 15 33 the the DET fis-4216 15 34 original original ADJ fis-4216 15 35 author author NOUN fis-4216 15 36 and and CCONJ fis-4216 15 37 source source NOUN fis-4216 15 38 are be AUX fis-4216 15 39 credited credit VERB fis-4216 15 40 . . PUNCT fis-4216 16 1 introduction introduction NOUN fis-4216 16 2 oncreted oncrete VERB fis-4216 16 3 civil civil ADJ fis-4216 16 4 infrastructure infrastructure NOUN fis-4216 16 5 such such ADJ fis-4216 16 6 as as ADP fis-4216 16 7 buildings building NOUN fis-4216 16 8 , , PUNCT fis-4216 16 9 bridges bridge NOUN fis-4216 16 10 , , PUNCT fis-4216 16 11 and and CCONJ fis-4216 16 12 highways highway NOUN fis-4216 16 13 sustain sustain VERB fis-4216 16 14 various various ADJ fis-4216 16 15 external external ADJ fis-4216 16 16 loads load NOUN fis-4216 16 17 throughout throughout ADP fis-4216 16 18 their their PRON fis-4216 16 19 service service NOUN fis-4216 16 20 period period NOUN fis-4216 16 21 . . PUNCT fis-4216 17 1 earthquake earthquake NOUN fis-4216 17 2 vibration vibration NOUN fis-4216 17 3 , , PUNCT fis-4216 17 4 wind wind NOUN fis-4216 17 5 , , PUNCT fis-4216 17 6 temperature temperature NOUN fis-4216 17 7 , , PUNCT fis-4216 17 8 or or CCONJ fis-4216 17 9 vehicle vehicle NOUN fis-4216 17 10 - - PUNCT fis-4216 17 11 made make VERB fis-4216 17 12 excitation excitation NOUN fis-4216 17 13 initiate initiate VERB fis-4216 17 14 structural structural ADJ fis-4216 17 15 damage damage NOUN fis-4216 17 16 during during ADP fis-4216 17 17 their their PRON fis-4216 17 18 lifetime lifetime NOUN fis-4216 17 19 and and CCONJ fis-4216 17 20 , , PUNCT fis-4216 17 21 subsequently subsequently ADV fis-4216 17 22 , , PUNCT fis-4216 17 23 trigger trigger NOUN fis-4216 17 24 catastrophic catastrophic ADJ fis-4216 17 25 failure failure NOUN fis-4216 17 26 [ [ X fis-4216 17 27 1 1 NUM fis-4216 17 28 , , PUNCT fis-4216 17 29 2 2 NUM fis-4216 17 30 ] ] PUNCT fis-4216 17 31 . . PUNCT fis-4216 18 1 regular regular ADJ fis-4216 18 2 inspection inspection NOUN fis-4216 18 3 and and CCONJ fis-4216 18 4 maintenance maintenance NOUN fis-4216 18 5 are be AUX fis-4216 18 6 vital vital ADJ fis-4216 18 7 to to ADP fis-4216 18 8 ensuring ensure VERB fis-4216 18 9 the the DET fis-4216 18 10 integrity integrity NOUN fis-4216 18 11 of of ADP fis-4216 18 12 concrete concrete ADJ fis-4216 18 13 infrastructure infrastructure NOUN fis-4216 18 14 for for ADP fis-4216 18 15 public public ADJ fis-4216 18 16 safety safety NOUN fis-4216 18 17 , , PUNCT fis-4216 18 18 economic economic ADJ fis-4216 18 19 running running NOUN fis-4216 18 20 , , PUNCT fis-4216 18 21 and and CCONJ fis-4216 18 22 national national ADJ fis-4216 18 23 security security NOUN fis-4216 18 24 . . PUNCT fis-4216 19 1 cracks crack NOUN fis-4216 19 2 on on ADP fis-4216 19 3 concrete concrete ADJ fis-4216 19 4 surfaces surface NOUN fis-4216 19 5 are be AUX fis-4216 19 6 essential essential ADJ fis-4216 19 7 indicators indicator NOUN fis-4216 19 8 of of ADP fis-4216 19 9 the the DET fis-4216 19 10 potential potential ADJ fis-4216 19 11 damage damage NOUN fis-4216 19 12 attached attach VERB fis-4216 19 13 to to ADP fis-4216 19 14 the the DET fis-4216 19 15 infrastructure infrastructure NOUN fis-4216 19 16 [ [ X fis-4216 19 17 1 1 NUM fis-4216 19 18 - - SYM fis-4216 19 19 5 5 NUM fis-4216 19 20 ] ] PUNCT fis-4216 19 21 . . PUNCT fis-4216 20 1 in in ADP fis-4216 20 2 this this DET fis-4216 20 3 regard regard NOUN fis-4216 20 4 , , PUNCT fis-4216 20 5 c c PROPN fis-4216 20 6 https://youtu.be/uy27wunphq4 https://youtu.be/uy27wunphq4 PROPN fis-4216 20 7 l. l. PROPN fis-4216 20 8 wang wang PROPN fis-4216 20 9 , , PUNCT fis-4216 20 10 frattura frattura PROPN fis-4216 20 11 ed ed PROPN fis-4216 20 12 integrità integrità PROPN fis-4216 20 13 strutturale strutturale PROPN fis-4216 20 14 , , PUNCT fis-4216 20 15 65 65 NUM fis-4216 20 16 ( ( PUNCT fis-4216 20 17 2023 2023 NUM fis-4216 20 18 ) ) PUNCT fis-4216 20 19 289 289 NUM fis-4216 20 20 - - SYM fis-4216 20 21 299 299 NUM fis-4216 20 22 ; ; PUNCT fis-4216 20 23 doi doi NOUN fis-4216 20 24 : : PUNCT fis-4216 20 25 10.3221 10.3221 NUM fis-4216 20 26 / / SYM fis-4216 20 27 igf igf PROPN fis-4216 20 28 - - PUNCT fis-4216 20 29 esis.65.19 esis.65.19 NOUN fis-4216 20 30 290 290 NUM fis-4216 20 31 detecting detect VERB fis-4216 20 32 cracks crack NOUN fis-4216 20 33 in in ADP fis-4216 20 34 structure structure NOUN fis-4216 20 35 surfaces surface NOUN fis-4216 20 36 in in ADP fis-4216 20 37 the the DET fis-4216 20 38 form form NOUN fis-4216 20 39 of of ADP fis-4216 20 40 digital digital ADJ fis-4216 20 41 images image NOUN fis-4216 20 42 captured capture VERB fis-4216 20 43 by by ADP fis-4216 20 44 unmanned unmanned ADJ fis-4216 20 45 aerial aerial ADJ fis-4216 20 46 vehicles vehicle NOUN fis-4216 20 47 ( ( PUNCT fis-4216 20 48 uavs uavs NOUN fis-4216 20 49 ) ) PUNCT fis-4216 21 1 [ [ X fis-4216 21 2 6 6 NUM fis-4216 21 3 , , PUNCT fis-4216 21 4 7 7 NUM fis-4216 21 5 ] ] PUNCT fis-4216 21 6 , , PUNCT fis-4216 21 7 smartphones smartphone NOUN fis-4216 21 8 [ [ X fis-4216 21 9 8 8 NUM fis-4216 21 10 , , PUNCT fis-4216 21 11 9 9 NUM fis-4216 21 12 ] ] PUNCT fis-4216 21 13 , , PUNCT fis-4216 21 14 or or CCONJ fis-4216 21 15 ground ground NOUN fis-4216 21 16 robots robot NOUN fis-4216 21 17 [ [ X fis-4216 21 18 10 10 NUM fis-4216 21 19 ] ] PUNCT fis-4216 21 20 has have AUX fis-4216 21 21 recently recently ADV fis-4216 21 22 gained gain VERB fis-4216 21 23 significant significant ADJ fis-4216 21 24 traction traction NOUN fis-4216 21 25 . . PUNCT fis-4216 22 1 the the DET fis-4216 22 2 first first ADJ fis-4216 22 3 and and CCONJ fis-4216 22 4 foremost foremost ADJ fis-4216 22 5 challenge challenge NOUN fis-4216 22 6 for for ADP fis-4216 22 7 a a DET fis-4216 22 8 structure structure NOUN fis-4216 22 9 inspection inspection NOUN fis-4216 22 10 system system NOUN fis-4216 22 11 is be AUX fis-4216 22 12 to to PART fis-4216 22 13 develop develop VERB fis-4216 22 14 a a DET fis-4216 22 15 computer computer NOUN fis-4216 22 16 model model NOUN fis-4216 22 17 that that PRON fis-4216 22 18 can can AUX fis-4216 22 19 automatically automatically ADV fis-4216 22 20 process process VERB fis-4216 22 21 the the DET fis-4216 22 22 image image NOUN fis-4216 22 23 data datum NOUN fis-4216 22 24 of of ADP fis-4216 22 25 concrete concrete ADJ fis-4216 22 26 surfaces surface NOUN fis-4216 22 27 and and CCONJ fis-4216 22 28 identify identify VERB fis-4216 22 29 the the DET fis-4216 22 30 signs sign NOUN fis-4216 22 31 of of ADP fis-4216 22 32 structural structural ADJ fis-4216 22 33 damage damage NOUN fis-4216 22 34 and and CCONJ fis-4216 22 35 distress distress NOUN fis-4216 22 36 [ [ X fis-4216 22 37 2–5 2–5 X fis-4216 22 38 ] ] PUNCT fis-4216 22 39 . . PUNCT fis-4216 23 1 therefore therefore ADV fis-4216 23 2 , , PUNCT fis-4216 23 3 researchers researcher NOUN fis-4216 23 4 developed develop VERB fis-4216 23 5 several several ADJ fis-4216 23 6 defect defect NOUN fis-4216 23 7 detection detection NOUN fis-4216 23 8 models model NOUN fis-4216 23 9 focusing focus VERB fis-4216 23 10 on on ADP fis-4216 23 11 structure structure NOUN fis-4216 23 12 crack crack NOUN fis-4216 23 13 detection detection NOUN fis-4216 23 14 . . PUNCT fis-4216 24 1 these these DET fis-4216 24 2 models model NOUN fis-4216 24 3 design design VERB fis-4216 24 4 a a DET fis-4216 24 5 variety variety NOUN fis-4216 24 6 of of ADP fis-4216 24 7 gradient gradient ADJ fis-4216 24 8 features feature NOUN fis-4216 24 9 or or CCONJ fis-4216 24 10 operators operator NOUN fis-4216 24 11 for for ADP fis-4216 24 12 each each DET fis-4216 24 13 image image NOUN fis-4216 24 14 pixel pixel NOUN fis-4216 24 15 and and CCONJ fis-4216 24 16 determine determine VERB fis-4216 24 17 whether whether SCONJ fis-4216 24 18 an an DET fis-4216 24 19 image image NOUN fis-4216 24 20 pixel pixel NOUN fis-4216 24 21 contains contain VERB fis-4216 24 22 a a DET fis-4216 24 23 crack crack NOUN fis-4216 24 24 with with ADP fis-4216 24 25 a a DET fis-4216 24 26 binary binary ADJ fis-4216 24 27 classifier classifier NOUN fis-4216 24 28 . . PUNCT fis-4216 25 1 however however ADV fis-4216 25 2 , , PUNCT fis-4216 25 3 these these DET fis-4216 25 4 models model NOUN fis-4216 25 5 highly highly ADV fis-4216 25 6 rely rely VERB fis-4216 25 7 on on ADP fis-4216 25 8 expert expert NOUN fis-4216 25 9 - - PUNCT fis-4216 25 10 driving drive VERB fis-4216 25 11 heuristic heuristic ADJ fis-4216 25 12 thresholds threshold NOUN fis-4216 25 13 or or CCONJ fis-4216 25 14 hand hand NOUN fis-4216 25 15 - - PUNCT fis-4216 25 16 crafted craft VERB fis-4216 25 17 features feature NOUN fis-4216 25 18 and and CCONJ fis-4216 25 19 are be AUX fis-4216 25 20 also also ADV fis-4216 25 21 not not PART fis-4216 25 22 discriminative discriminative VERB fis-4216 25 23 enough enough ADV fis-4216 25 24 to to PART fis-4216 25 25 differentiate differentiate VERB fis-4216 25 26 the the DET fis-4216 25 27 crack crack NOUN fis-4216 25 28 from from ADP fis-4216 25 29 complex complex ADJ fis-4216 25 30 image image NOUN fis-4216 25 31 variations variation NOUN fis-4216 25 32 . . PUNCT fis-4216 26 1 convolutional convolutional ADJ fis-4216 26 2 neural neural ADJ fis-4216 26 3 networks network NOUN fis-4216 26 4 ( ( PUNCT fis-4216 26 5 cnns)based cnns)base VERB fis-4216 26 6 deep deep ADJ fis-4216 26 7 learning learning NOUN fis-4216 26 8 becomes become VERB fis-4216 26 9 an an DET fis-4216 26 10 alternate alternate ADJ fis-4216 26 11 method method NOUN fis-4216 26 12 for for ADP fis-4216 26 13 automatic automatic ADJ fis-4216 26 14 crack crack NOUN fis-4216 26 15 detection detection NOUN fis-4216 26 16 , , PUNCT fis-4216 26 17 inspired inspire VERB fis-4216 26 18 by by ADP fis-4216 26 19 successful successful ADJ fis-4216 26 20 applications application NOUN fis-4216 26 21 in in ADP fis-4216 26 22 medical medical ADJ fis-4216 26 23 diagnosis diagnosis NOUN fis-4216 26 24 . . PUNCT fis-4216 27 1 instead instead ADV fis-4216 27 2 of of ADP fis-4216 27 3 the the DET fis-4216 27 4 hand hand NOUN fis-4216 27 5 - - PUNCT fis-4216 27 6 crafted craft VERB fis-4216 27 7 features feature NOUN fis-4216 27 8 , , PUNCT fis-4216 27 9 the the DET fis-4216 27 10 cnns cnn NOUN fis-4216 27 11 can can AUX fis-4216 27 12 automatically automatically ADV fis-4216 27 13 learn learn VERB fis-4216 27 14 the the DET fis-4216 27 15 discriminative discriminative NOUN fis-4216 27 16 features feature VERB fis-4216 27 17 from from ADP fis-4216 27 18 the the DET fis-4216 27 19 digital digital ADJ fis-4216 27 20 image image NOUN fis-4216 27 21 to to PART fis-4216 27 22 decide decide VERB fis-4216 27 23 whether whether SCONJ fis-4216 27 24 a a DET fis-4216 27 25 crack crack NOUN fis-4216 27 26 exists exist VERB fis-4216 27 27 . . PUNCT fis-4216 28 1 zhang zhang PROPN fis-4216 28 2 et et PROPN fis-4216 28 3 al al PROPN fis-4216 28 4 . . PROPN fis-4216 28 5 developed develop VERB fis-4216 28 6 a a DET fis-4216 28 7 cnn cnn PROPN fis-4216 28 8 model model NOUN fis-4216 28 9 and and CCONJ fis-4216 28 10 trained train VERB fis-4216 28 11 it it PRON fis-4216 28 12 using use VERB fis-4216 28 13 image image NOUN fis-4216 28 14 patches patch NOUN fis-4216 28 15 taken take VERB fis-4216 28 16 from from ADP fis-4216 28 17 pavement pavement NOUN fis-4216 28 18 [ [ X fis-4216 28 19 11 11 NUM fis-4216 28 20 ] ] PUNCT fis-4216 28 21 . . PUNCT fis-4216 29 1 a a DET fis-4216 29 2 notable notable ADJ fis-4216 29 3 increase increase NOUN fis-4216 29 4 in in ADP fis-4216 29 5 prediction prediction NOUN fis-4216 29 6 accuracy accuracy NOUN fis-4216 29 7 was be AUX fis-4216 29 8 achieved achieve VERB fis-4216 29 9 compared compare VERB fis-4216 29 10 to to PART fis-4216 29 11 support support VERB fis-4216 29 12 vector vector NOUN fis-4216 29 13 machines machine NOUN fis-4216 29 14 and and CCONJ fis-4216 29 15 boosting boost VERB fis-4216 29 16 methods method NOUN fis-4216 29 17 . . PUNCT fis-4216 30 1 cha cha PROPN fis-4216 30 2 et et PROPN fis-4216 30 3 al al PROPN fis-4216 30 4 . . PROPN fis-4216 30 5 established establish VERB fis-4216 30 6 an an DET fis-4216 30 7 image image NOUN fis-4216 30 8 dataset dataset NOUN fis-4216 30 9 consisting consist VERB fis-4216 30 10 of of ADP fis-4216 30 11 a a DET fis-4216 30 12 total total NOUN fis-4216 30 13 of of ADP fis-4216 30 14 2,366 2,366 NUM fis-4216 30 15 sub sub NOUN fis-4216 30 16 - - ADJ fis-4216 30 17 images image NOUN fis-4216 30 18 cropped crop VERB fis-4216 30 19 from from ADP fis-4216 30 20 300 300 NUM fis-4216 30 21 annotated annotate VERB fis-4216 30 22 images image NOUN fis-4216 30 23 of of ADP fis-4216 30 24 two two NUM fis-4216 30 25 bridges bridge NOUN fis-4216 30 26 and and CCONJ fis-4216 30 27 a a DET fis-4216 30 28 building building NOUN fis-4216 30 29 [ [ X fis-4216 30 30 12 12 NUM fis-4216 30 31 ] ] PUNCT fis-4216 30 32 . . PUNCT fis-4216 31 1 he he PRON fis-4216 31 2 subsequently subsequently ADV fis-4216 31 3 trained train VERB fis-4216 31 4 a a DET fis-4216 31 5 faster fast ADJ fis-4216 31 6 region region NOUN fis-4216 31 7 - - PUNCT fis-4216 31 8 based base VERB fis-4216 31 9 cnn cnn NOUN fis-4216 31 10 architecture architecture NOUN fis-4216 31 11 on on ADP fis-4216 31 12 this this DET fis-4216 31 13 image image NOUN fis-4216 31 14 dataset dataset VERB fis-4216 31 15 to to PART fis-4216 31 16 identify identify VERB fis-4216 31 17 damage damage NOUN fis-4216 31 18 with with ADP fis-4216 31 19 a a DET fis-4216 31 20 reported report VERB fis-4216 31 21 average average ADJ fis-4216 31 22 precision precision NOUN fis-4216 31 23 of of ADP fis-4216 31 24 88 88 NUM fis-4216 31 25 % % NOUN fis-4216 31 26 . . PUNCT fis-4216 32 1 alipour alipour PROPN fis-4216 32 2 et et PROPN fis-4216 32 3 al al PROPN fis-4216 32 4 . . PROPN fis-4216 32 5 developed develop VERB fis-4216 32 6 a a DET fis-4216 32 7 full full ADJ fis-4216 32 8 convolutional convolutional ADJ fis-4216 32 9 network network NOUN fis-4216 32 10 fcn2s fcn2s PROPN fis-4216 32 11 for for ADP fis-4216 32 12 pixel pixel ADJ fis-4216 32 13 - - PUNCT fis-4216 32 14 level level NOUN fis-4216 32 15 defect defect NOUN fis-4216 32 16 detection detection NOUN fis-4216 32 17 in in ADP fis-4216 32 18 concrete concrete ADJ fis-4216 32 19 infrastructure infrastructure NOUN fis-4216 32 20 systems system NOUN fis-4216 32 21 . . PUNCT fis-4216 33 1 sensitive sensitive ADJ fis-4216 33 2 analysis analysis NOUN fis-4216 33 3 revealed reveal VERB fis-4216 33 4 that that SCONJ fis-4216 33 5 their their PRON fis-4216 33 6 model model NOUN fis-4216 33 7 could could AUX fis-4216 33 8 correctly correctly ADV fis-4216 33 9 detect detect VERB fis-4216 33 10 over over ADP fis-4216 33 11 92 92 NUM fis-4216 33 12 % % NOUN fis-4216 33 13 of of ADP fis-4216 33 14 crack crack NOUN fis-4216 33 15 pixels pixel NOUN fis-4216 33 16 [ [ X fis-4216 33 17 4 4 NUM fis-4216 33 18 ] ] PUNCT fis-4216 33 19 . . PUNCT fis-4216 34 1 chen chen PROPN fis-4216 34 2 and and CCONJ fis-4216 34 3 he he PRON fis-4216 34 4 present present VERB fis-4216 34 5 a a DET fis-4216 34 6 novel novel ADJ fis-4216 34 7 u u ADJ fis-4216 34 8 - - ADJ fis-4216 34 9 shaped shape VERB fis-4216 34 10 encoder encoder NOUN fis-4216 34 11 - - PUNCT fis-4216 34 12 decoder decoder NOUN fis-4216 34 13 network network NOUN fis-4216 34 14 with with ADP fis-4216 34 15 an an DET fis-4216 34 16 attention attention NOUN fis-4216 34 17 mechanism mechanism NOUN fis-4216 34 18 to to PART fis-4216 34 19 achieve achieve VERB fis-4216 34 20 enhanced enhanced ADJ fis-4216 34 21 accuracy accuracy NOUN fis-4216 34 22 in in ADP fis-4216 34 23 pixel pixel ADJ fis-4216 34 24 - - PUNCT fis-4216 34 25 level level NOUN fis-4216 34 26 crack crack NOUN fis-4216 34 27 detection detection NOUN fis-4216 34 28 of of ADP fis-4216 34 29 concrete concrete ADJ fis-4216 34 30 roads road NOUN fis-4216 34 31 compared compare VERB fis-4216 34 32 with with ADP fis-4216 34 33 other other ADJ fis-4216 34 34 advanced advanced ADJ fis-4216 34 35 networks network NOUN fis-4216 34 36 [ [ X fis-4216 34 37 13 13 NUM fis-4216 34 38 ] ] PUNCT fis-4216 34 39 . . PUNCT fis-4216 35 1 zhu zhu PROPN fis-4216 35 2 et et PROPN fis-4216 35 3 al al PROPN fis-4216 35 4 . . PROPN fis-4216 35 5 compared compare VERB fis-4216 35 6 the the DET fis-4216 35 7 performance performance NOUN fis-4216 35 8 of of ADP fis-4216 35 9 the the DET fis-4216 35 10 three three NUM fis-4216 35 11 cnn cnn PROPN fis-4216 35 12 algorithms algorithm NOUN fis-4216 35 13 - - PUNCT fis-4216 35 14 fast fast ADJ fis-4216 35 15 r r NOUN fis-4216 35 16 - - PUNCT fis-4216 35 17 cnn cnn PROPN fis-4216 35 18 , , PUNCT fis-4216 35 19 yolov3 yolov3 PROPN fis-4216 35 20 , , PUNCT fis-4216 35 21 and and CCONJ fis-4216 35 22 yolov4 yolov4 NOUN fis-4216 35 23 - - PUNCT fis-4216 35 24 for for ADP fis-4216 35 25 distress distress NOUN fis-4216 35 26 detection detection NOUN fis-4216 35 27 based base VERB fis-4216 35 28 on on ADP fis-4216 35 29 uav uav PROPN fis-4216 35 30 - - PUNCT fis-4216 35 31 captured capture VERB fis-4216 35 32 pavement pavement NOUN fis-4216 35 33 images image NOUN fis-4216 35 34 [ [ X fis-4216 35 35 14 14 NUM fis-4216 35 36 ] ] PUNCT fis-4216 35 37 . . PUNCT fis-4216 36 1 shang shang PROPN fis-4216 36 2 et et PROPN fis-4216 36 3 al al PROPN fis-4216 36 4 . . PROPN fis-4216 36 5 presented present VERB fis-4216 36 6 a a DET fis-4216 36 7 multi multi ADJ fis-4216 36 8 - - ADJ fis-4216 36 9 fusion fusion ADJ fis-4216 36 10 u u ADJ fis-4216 36 11 - - ADJ fis-4216 36 12 net net ADJ fis-4216 36 13 network network NOUN fis-4216 36 14 to to PART fis-4216 36 15 automatically automatically ADV fis-4216 36 16 detect detect VERB fis-4216 36 17 the the DET fis-4216 36 18 pixel pixel ADJ fis-4216 36 19 - - PUNCT fis-4216 36 20 level level NOUN fis-4216 36 21 pavement pavement NOUN fis-4216 36 22 crack crack NOUN fis-4216 36 23 with with ADP fis-4216 36 24 superior superior ADJ fis-4216 36 25 performance performance NOUN fis-4216 36 26 compared compare VERB fis-4216 36 27 to to ADP fis-4216 36 28 seven seven NUM fis-4216 36 29 other other ADJ fis-4216 36 30 state state NOUN fis-4216 36 31 - - PUNCT fis-4216 36 32 of of ADP fis-4216 36 33 - - PUNCT fis-4216 36 34 the the DET fis-4216 36 35 - - PUNCT fis-4216 36 36 art art NOUN fis-4216 36 37 models model NOUN fis-4216 36 38 [ [ X fis-4216 36 39 15 15 NUM fis-4216 36 40 ] ] PUNCT fis-4216 36 41 . . PUNCT fis-4216 37 1 recently recently ADV fis-4216 37 2 , , PUNCT fis-4216 37 3 ha ha INTJ fis-4216 37 4 et et PROPN fis-4216 37 5 al al PROPN fis-4216 37 6 . . PUNCT fis-4216 37 7 develop develop VERB fis-4216 37 8 a a DET fis-4216 37 9 novel novel ADJ fis-4216 37 10 system system NOUN fis-4216 37 11 integrating integrate VERB fis-4216 37 12 squeezenet squeezenet NOUN fis-4216 37 13 , , PUNCT fis-4216 37 14 u u NOUN fis-4216 37 15 - - NOUN fis-4216 37 16 net net ADJ fis-4216 37 17 and and CCONJ fis-4216 37 18 mobilenet mobilenet NOUN fis-4216 37 19 - - PUNCT fis-4216 37 20 ssd ssd NOUN fis-4216 37 21 models model NOUN fis-4216 37 22 together together ADV fis-4216 37 23 for for ADP fis-4216 37 24 detection detection NOUN fis-4216 37 25 , , PUNCT fis-4216 37 26 classification classification NOUN fis-4216 37 27 , , PUNCT fis-4216 37 28 and and CCONJ fis-4216 37 29 severity severity NOUN fis-4216 37 30 assessment assessment NOUN fis-4216 37 31 of of ADP fis-4216 37 32 five five NUM fis-4216 37 33 types type NOUN fis-4216 37 34 road road NOUN fis-4216 37 35 cracks crack NOUN fis-4216 37 36 [ [ X fis-4216 37 37 16 16 NUM fis-4216 37 38 ] ] PUNCT fis-4216 37 39 . . PUNCT fis-4216 38 1 this this DET fis-4216 38 2 combined combined ADJ fis-4216 38 3 system system NOUN fis-4216 38 4 is be AUX fis-4216 38 5 capable capable ADJ fis-4216 38 6 of of ADP fis-4216 38 7 crack crack NOUN fis-4216 38 8 severity severity NOUN fis-4216 38 9 assessment assessment NOUN fis-4216 38 10 and and CCONJ fis-4216 38 11 crack crack VERB fis-4216 38 12 type type NOUN fis-4216 38 13 classification classification NOUN fis-4216 38 14 with with ADP fis-4216 38 15 a a DET fis-4216 38 16 reported report VERB fis-4216 38 17 accuracy accuracy NOUN fis-4216 38 18 of of ADP fis-4216 38 19 91.2 91.2 NUM fis-4216 38 20 % % NOUN fis-4216 38 21 . . PUNCT fis-4216 39 1 the the DET fis-4216 39 2 recent recent ADJ fis-4216 39 3 success success NOUN fis-4216 39 4 of of ADP fis-4216 39 5 deep deep ADJ fis-4216 39 6 learning learning NOUN fis-4216 39 7 methods method NOUN fis-4216 39 8 for for ADP fis-4216 39 9 crack crack NOUN fis-4216 39 10 detection detection NOUN fis-4216 39 11 applications application NOUN fis-4216 39 12 demonstrates demonstrate VERB fis-4216 39 13 the the DET fis-4216 39 14 feasibility feasibility NOUN fis-4216 39 15 of of ADP fis-4216 39 16 fine fine ADV fis-4216 39 17 - - PUNCT fis-4216 39 18 tuning tune VERB fis-4216 39 19 cnns cnn NOUN fis-4216 39 20 within within ADP fis-4216 39 21 an an DET fis-4216 39 22 ample ample ADJ fis-4216 39 23 parameter parameter NOUN fis-4216 39 24 space space NOUN fis-4216 40 1 [ [ X fis-4216 40 2 2 2 NUM fis-4216 40 3 - - SYM fis-4216 40 4 5 5 NUM fis-4216 40 5 , , PUNCT fis-4216 40 6 11 11 NUM fis-4216 40 7 - - SYM fis-4216 40 8 16 16 NUM fis-4216 40 9 ] ] PUNCT fis-4216 40 10 . . PUNCT fis-4216 41 1 for for ADP fis-4216 41 2 instance instance NOUN fis-4216 41 3 , , PUNCT fis-4216 41 4 the the DET fis-4216 41 5 cnn cnn PROPN fis-4216 41 6 models model NOUN fis-4216 41 7 , , PUNCT fis-4216 41 8 such such ADJ fis-4216 41 9 as as ADP fis-4216 41 10 convnet convnet NOUN fis-4216 41 11 , , PUNCT fis-4216 41 12 u u NOUN fis-4216 41 13 - - NOUN fis-4216 41 14 net net ADJ fis-4216 41 15 , , PUNCT fis-4216 41 16 and and CCONJ fis-4216 41 17 yolo yolo PROPN fis-4216 41 18 adopted adopt VERB fis-4216 41 19 in in ADP fis-4216 41 20 [ [ X fis-4216 41 21 11 11 NUM fis-4216 41 22 , , PUNCT fis-4216 41 23 14 14 NUM fis-4216 41 24 , , PUNCT fis-4216 41 25 15 15 NUM fis-4216 41 26 , , PUNCT fis-4216 41 27 16 16 NUM fis-4216 41 28 ] ] PUNCT fis-4216 41 29 , , PUNCT fis-4216 41 30 have have VERB fis-4216 41 31 more more ADJ fis-4216 41 32 than than ADP fis-4216 41 33 60 60 NUM fis-4216 41 34 million million NUM fis-4216 41 35 parameters parameter NOUN fis-4216 41 36 . . PUNCT fis-4216 42 1 the the DET fis-4216 42 2 fcn2s fcn2s PROPN fis-4216 42 3 network network NOUN fis-4216 42 4 for for ADP fis-4216 42 5 pixel pixel ADJ fis-4216 42 6 - - PUNCT fis-4216 42 7 level level NOUN fis-4216 42 8 crack crack NOUN fis-4216 42 9 identification identification NOUN fis-4216 42 10 has have VERB fis-4216 42 11 144 144 NUM fis-4216 42 12 million million NUM fis-4216 42 13 parameters parameter NOUN fis-4216 42 14 [ [ X fis-4216 42 15 4 4 NUM fis-4216 42 16 ] ] PUNCT fis-4216 42 17 . . PUNCT fis-4216 43 1 iterative iterative ADJ fis-4216 43 2 optimization optimization NOUN fis-4216 43 3 of of ADP fis-4216 43 4 such such DET fis-4216 43 5 a a DET fis-4216 43 6 vast vast ADJ fis-4216 43 7 parameter parameter NOUN fis-4216 43 8 space space NOUN fis-4216 43 9 to to PART fis-4216 43 10 determine determine VERB fis-4216 43 11 the the DET fis-4216 43 12 best good ADJ fis-4216 43 13 network network NOUN fis-4216 43 14 model model NOUN fis-4216 43 15 is be AUX fis-4216 43 16 time time NOUN fis-4216 43 17 - - PUNCT fis-4216 43 18 consuming consume VERB fis-4216 43 19 and and CCONJ fis-4216 43 20 computationally computationally ADV fis-4216 43 21 costly costly ADJ fis-4216 43 22 . . PUNCT fis-4216 44 1 it it PRON fis-4216 44 2 is be AUX fis-4216 44 3 eager eager ADJ fis-4216 44 4 to to PART fis-4216 44 5 develop develop VERB fis-4216 44 6 a a DET fis-4216 44 7 deep deep ADJ fis-4216 44 8 learning learning NOUN fis-4216 44 9 - - PUNCT fis-4216 44 10 based base VERB fis-4216 44 11 model model NOUN fis-4216 44 12 for for ADP fis-4216 44 13 automated automate VERB fis-4216 44 14 crack crack NOUN fis-4216 44 15 detection detection NOUN fis-4216 44 16 with with ADP fis-4216 44 17 a a DET fis-4216 44 18 compact compact ADJ fis-4216 44 19 network network NOUN fis-4216 44 20 structure structure NOUN fis-4216 44 21 for for ADP fis-4216 44 22 easy easy ADJ fis-4216 44 23 system system NOUN fis-4216 44 24 embedding embed VERB fis-4216 44 25 while while SCONJ fis-4216 44 26 maintaining maintain VERB fis-4216 44 27 equivalent equivalent ADJ fis-4216 44 28 or or CCONJ fis-4216 44 29 even even ADV fis-4216 44 30 better well ADJ fis-4216 44 31 performance performance NOUN fis-4216 44 32 . . PUNCT fis-4216 45 1 to to ADP fis-4216 45 2 this this DET fis-4216 45 3 end end NOUN fis-4216 45 4 , , PUNCT fis-4216 45 5 a a DET fis-4216 45 6 deep deep ADJ fis-4216 45 7 learning learning NOUN fis-4216 45 8 - - PUNCT fis-4216 45 9 based base VERB fis-4216 45 10 system system NOUN fis-4216 45 11 called call VERB fis-4216 45 12 cracksn cracksn NOUN fis-4216 45 13 is be AUX fis-4216 45 14 implemented implement VERB fis-4216 45 15 for for ADP fis-4216 45 16 automatic automatic ADJ fis-4216 45 17 crack crack NOUN fis-4216 45 18 detection detection NOUN fis-4216 45 19 in in ADP fis-4216 45 20 concrete concrete ADJ fis-4216 45 21 infrastructure infrastructure NOUN fis-4216 45 22 , , PUNCT fis-4216 45 23 integrating integrate VERB fis-4216 45 24 the the DET fis-4216 45 25 squeezenet squeezenet NOUN fis-4216 45 26 network network NOUN fis-4216 45 27 and and CCONJ fis-4216 45 28 the the DET fis-4216 45 29 adam adam PROPN fis-4216 45 30 optimization optimization NOUN fis-4216 45 31 algorithm algorithm NOUN fis-4216 45 32 . . PUNCT fis-4216 46 1 the the DET fis-4216 46 2 proposed propose VERB fis-4216 46 3 deep deep ADJ fis-4216 46 4 learning learning NOUN fis-4216 46 5 model model NOUN fis-4216 46 6 , , PUNCT fis-4216 46 7 which which PRON fis-4216 46 8 enables enable VERB fis-4216 46 9 fast fast ADJ fis-4216 46 10 and and CCONJ fis-4216 46 11 stable stable ADJ fis-4216 46 12 inspection inspection NOUN fis-4216 46 13 of of ADP fis-4216 46 14 concrete concrete ADJ fis-4216 46 15 cracks crack NOUN fis-4216 46 16 , , PUNCT fis-4216 46 17 has have VERB fis-4216 46 18 a a DET fis-4216 46 19 lightweight lightweight ADJ fis-4216 46 20 structure structure NOUN fis-4216 46 21 architecture architecture NOUN fis-4216 46 22 compared compare VERB fis-4216 46 23 to to ADP fis-4216 46 24 other other ADJ fis-4216 46 25 cnns cnn NOUN fis-4216 46 26 and and CCONJ fis-4216 46 27 is be AUX fis-4216 46 28 , , PUNCT fis-4216 46 29 therefore therefore ADV fis-4216 46 30 , , PUNCT fis-4216 46 31 easy easy ADJ fis-4216 46 32 to to PART fis-4216 46 33 implement implement VERB fis-4216 46 34 for for ADP fis-4216 46 35 embedded embed VERB fis-4216 46 36 and and CCONJ fis-4216 46 37 mobile mobile NOUN fis-4216 46 38 systems system NOUN fis-4216 46 39 . . PUNCT fis-4216 47 1 the the DET fis-4216 47 2 proposed propose VERB fis-4216 47 3 deep deep ADJ fis-4216 47 4 learning learning NOUN fis-4216 47 5 adamsqueezenet adamsqueezenet NOUN fis-4216 47 6 based base VERB fis-4216 47 7 system system NOUN fis-4216 47 8 , , PUNCT fis-4216 47 9 and and CCONJ fis-4216 47 10 components component NOUN fis-4216 47 11 , , PUNCT fis-4216 47 12 i.e. i.e. X fis-4216 47 13 , , PUNCT fis-4216 47 14 image image NOUN fis-4216 47 15 data datum NOUN fis-4216 47 16 preparation preparation NOUN fis-4216 47 17 , , PUNCT fis-4216 47 18 squeezenet squeezenet NOUN fis-4216 47 19 architecture architecture NOUN fis-4216 47 20 , , PUNCT fis-4216 47 21 and and CCONJ fis-4216 47 22 adam adam PROPN fis-4216 47 23 optimization optimization NOUN fis-4216 47 24 , , PUNCT fis-4216 47 25 are be AUX fis-4216 47 26 described describe VERB fis-4216 47 27 in in ADP fis-4216 47 28 detail detail NOUN fis-4216 47 29 . . PUNCT fis-4216 48 1 the the DET fis-4216 48 2 experimental experimental ADJ fis-4216 48 3 results result NOUN fis-4216 48 4 of of ADP fis-4216 48 5 the the DET fis-4216 48 6 presented present VERB fis-4216 48 7 model model NOUN fis-4216 48 8 demonstrate demonstrate VERB fis-4216 48 9 the the DET fis-4216 48 10 capability capability NOUN fis-4216 48 11 of of ADP fis-4216 48 12 the the DET fis-4216 48 13 cracksn cracksn NOUN fis-4216 48 14 system system NOUN fis-4216 48 15 for for ADP fis-4216 48 16 crack crack NOUN fis-4216 48 17 detection detection NOUN fis-4216 48 18 . . PUNCT fis-4216 49 1 the the DET fis-4216 49 2 presented present VERB fis-4216 49 3 system system NOUN fis-4216 49 4 's 's PART fis-4216 49 5 limitations limitation NOUN fis-4216 49 6 are be AUX fis-4216 49 7 discussed discuss VERB fis-4216 49 8 , , PUNCT fis-4216 49 9 followed follow VERB fis-4216 49 10 by by ADP fis-4216 49 11 a a DET fis-4216 49 12 brief brief ADJ fis-4216 49 13 conclusion conclusion NOUN fis-4216 49 14 . . PUNCT fis-4216 50 1 methods method NOUN fis-4216 50 2 ue ue INTJ fis-4216 50 3 to to ADP fis-4216 50 4 their their PRON fis-4216 50 5 strong strong ADJ fis-4216 50 6 self self NOUN fis-4216 50 7 - - PUNCT fis-4216 50 8 learning learn VERB fis-4216 50 9 abilities ability NOUN fis-4216 50 10 , , PUNCT fis-4216 50 11 cnn cnn PROPN fis-4216 50 12 models model NOUN fis-4216 50 13 can can AUX fis-4216 50 14 classify classify VERB fis-4216 50 15 large large ADJ fis-4216 50 16 - - PUNCT fis-4216 50 17 scale scale NOUN fis-4216 50 18 image image NOUN fis-4216 50 19 datasets dataset NOUN fis-4216 50 20 with with ADP fis-4216 50 21 human human NOUN fis-4216 50 22 - - PUNCT fis-4216 50 23 like like ADJ fis-4216 50 24 accuracies accuracy NOUN fis-4216 50 25 [ [ X fis-4216 50 26 17 17 NUM fis-4216 50 27 ] ] PUNCT fis-4216 50 28 . . PUNCT fis-4216 51 1 a a DET fis-4216 51 2 convolutional convolutional ADJ fis-4216 51 3 neural neural ADJ fis-4216 51 4 network network NOUN fis-4216 51 5 is be AUX fis-4216 51 6 a a DET fis-4216 51 7 chain chain NOUN fis-4216 51 8 of of ADP fis-4216 51 9 collaborative collaborative ADJ fis-4216 51 10 convolution convolution NOUN fis-4216 51 11 layers layer NOUN fis-4216 51 12 , , PUNCT fis-4216 51 13 activation activation NOUN fis-4216 51 14 functions function NOUN fis-4216 51 15 , , PUNCT fis-4216 51 16 pooling pooling NOUN fis-4216 51 17 , , PUNCT fis-4216 51 18 and and CCONJ fis-4216 51 19 batch batch VERB fis-4216 51 20 normalization normalization NOUN fis-4216 51 21 operations operation NOUN fis-4216 51 22 . . PUNCT fis-4216 52 1 sequential sequential ADJ fis-4216 52 2 convolution convolution NOUN fis-4216 52 3 layers layer NOUN fis-4216 52 4 transform transform VERB fis-4216 52 5 the the DET fis-4216 52 6 2d 2d NOUN fis-4216 52 7 or or CCONJ fis-4216 52 8 3d 3d NUM fis-4216 52 9 image image NOUN fis-4216 52 10 inputs input NOUN fis-4216 52 11 into into ADP fis-4216 52 12 a a DET fis-4216 52 13 high high ADJ fis-4216 52 14 - - PUNCT fis-4216 52 15 level level NOUN fis-4216 52 16 feature feature NOUN fis-4216 52 17 map map NOUN fis-4216 52 18 with with ADP fis-4216 52 19 specified specified ADJ fis-4216 52 20 kernels kernel NOUN fis-4216 52 21 , , PUNCT fis-4216 52 22 significantly significantly ADV fis-4216 52 23 reducing reduce VERB fis-4216 52 24 computation computation NOUN fis-4216 52 25 complexity complexity NOUN fis-4216 52 26 and and CCONJ fis-4216 52 27 enhancing enhance VERB fis-4216 52 28 generalization generalization NOUN fis-4216 52 29 ability ability NOUN fis-4216 52 30 [ [ X fis-4216 52 31 18 18 NUM fis-4216 52 32 ] ] PUNCT fis-4216 52 33 . . PUNCT fis-4216 53 1 in in ADP fis-4216 53 2 this this DET fis-4216 53 3 work work NOUN fis-4216 53 4 , , PUNCT fis-4216 53 5 we we PRON fis-4216 53 6 propose propose VERB fis-4216 53 7 a a DET fis-4216 53 8 system system NOUN fis-4216 53 9 , , PUNCT fis-4216 53 10 namely namely ADV fis-4216 53 11 cracksn cracksn NOUN fis-4216 53 12 , , PUNCT fis-4216 53 13 integrated integrate VERB fis-4216 53 14 adams adams PROPN fis-4216 53 15 optimization optimization NOUN fis-4216 53 16 , , PUNCT fis-4216 53 17 and and CCONJ fis-4216 53 18 squeezenet squeezenet NOUN fis-4216 53 19 architecture architecture NOUN fis-4216 53 20 , , PUNCT fis-4216 53 21 for for SCONJ fis-4216 53 22 the the DET fis-4216 53 23 diagnostic diagnostic NOUN fis-4216 53 24 of of ADP fis-4216 53 25 concrete concrete ADJ fis-4216 53 26 cracks crack NOUN fis-4216 53 27 from from ADP fis-4216 53 28 digital digital ADJ fis-4216 53 29 images image NOUN fis-4216 53 30 is be AUX fis-4216 53 31 proposed propose VERB fis-4216 53 32 in in ADP fis-4216 53 33 this this DET fis-4216 53 34 work work NOUN fis-4216 53 35 . . PUNCT fis-4216 54 1 the the DET fis-4216 54 2 proposed propose VERB fis-4216 54 3 framework framework NOUN fis-4216 54 4 intends intend VERB fis-4216 54 5 to to PART fis-4216 54 6 provide provide VERB fis-4216 54 7 a a DET fis-4216 54 8 discriminative discriminative NOUN fis-4216 54 9 and and CCONJ fis-4216 54 10 rapid rapid ADJ fis-4216 54 11 system system NOUN fis-4216 54 12 for for ADP fis-4216 54 13 automatically automatically ADV fis-4216 54 14 detecting detect VERB fis-4216 54 15 distress distress NOUN fis-4216 54 16 in in ADP fis-4216 54 17 concrete concrete ADJ fis-4216 54 18 surfaces surface NOUN fis-4216 54 19 from from ADP fis-4216 54 20 digital digital ADJ fis-4216 54 21 images image NOUN fis-4216 54 22 , , PUNCT fis-4216 54 23 i.e. i.e. X fis-4216 54 24 , , PUNCT fis-4216 54 25 captured capture VERB fis-4216 54 26 by by ADP fis-4216 54 27 smartphones smartphone NOUN fis-4216 54 28 or or CCONJ fis-4216 54 29 uavs uavs NOUN fis-4216 54 30 . . PUNCT fis-4216 55 1 this this DET fis-4216 55 2 system system NOUN fis-4216 55 3 can can AUX fis-4216 55 4 also also ADV fis-4216 55 5 facilitate facilitate VERB fis-4216 55 6 quick quick ADJ fis-4216 55 7 status status NOUN fis-4216 55 8 evaluation evaluation NOUN fis-4216 55 9 and and CCONJ fis-4216 55 10 decision decision NOUN fis-4216 55 11 - - PUNCT fis-4216 55 12 making making NOUN fis-4216 55 13 in in ADP fis-4216 55 14 the the DET fis-4216 55 15 maintenance maintenance NOUN fis-4216 55 16 of of ADP fis-4216 55 17 concrete concrete ADJ fis-4216 55 18 infrastructure infrastructure NOUN fis-4216 55 19 . . PUNCT fis-4216 56 1 the the DET fis-4216 56 2 following follow VERB fis-4216 56 3 sections section NOUN fis-4216 56 4 describe describe VERB fis-4216 56 5 the the DET fis-4216 56 6 proposed propose VERB fis-4216 56 7 framework framework NOUN fis-4216 56 8 , , PUNCT fis-4216 56 9 image image NOUN fis-4216 56 10 data datum NOUN fis-4216 56 11 preparation preparation NOUN fis-4216 56 12 , , PUNCT fis-4216 56 13 squeezenet squeezenet NOUN fis-4216 56 14 architecture architecture NOUN fis-4216 56 15 , , PUNCT fis-4216 56 16 and and CCONJ fis-4216 56 17 adam adam PROPN fis-4216 56 18 optimization optimization NOUN fis-4216 56 19 in in ADP fis-4216 56 20 detail detail NOUN fis-4216 56 21 . . PUNCT fis-4216 57 1 d d PROPN fis-4216 57 2 l. l. PROPN fis-4216 57 3 wang wang PROPN fis-4216 57 4 , , PUNCT fis-4216 57 5 frattura frattura PROPN fis-4216 57 6 ed ed PROPN fis-4216 57 7 integrità integrità PROPN fis-4216 57 8 strutturale strutturale PROPN fis-4216 57 9 , , PUNCT fis-4216 57 10 65 65 NUM fis-4216 57 11 ( ( PUNCT fis-4216 57 12 2023 2023 NUM fis-4216 57 13 ) ) PUNCT fis-4216 57 14 289 289 NUM fis-4216 57 15 - - SYM fis-4216 57 16 299 299 NUM fis-4216 57 17 ; ; PUNCT fis-4216 57 18 doi doi NOUN fis-4216 57 19 : : PUNCT fis-4216 57 20 10.3221 10.3221 NUM fis-4216 57 21 / / SYM fis-4216 57 22 igf igf PROPN fis-4216 57 23 - - PUNCT fis-4216 57 24 esis.65.19 esis.65.19 NOUN fis-4216 57 25 291 291 NUM fis-4216 57 26 proposed propose VERB fis-4216 57 27 system system NOUN fis-4216 57 28 : : PUNCT fis-4216 57 29 adam adam NOUN fis-4216 57 30 - - PUNCT fis-4216 57 31 squeezenet squeezenet NOUN fis-4216 57 32 based base VERB fis-4216 57 33 cracksn cracksn PROPN fis-4216 57 34 fig fig NOUN fis-4216 57 35 . . PUNCT fis-4216 58 1 1 1 NUM fis-4216 58 2 describes describe VERB fis-4216 58 3 the the DET fis-4216 58 4 overall overall ADJ fis-4216 58 5 architecture architecture NOUN fis-4216 58 6 of of ADP fis-4216 58 7 the the DET fis-4216 58 8 adam adam NOUN fis-4216 58 9 - - PUNCT fis-4216 58 10 squeezenet squeezenet NOUN fis-4216 58 11 based base VERB fis-4216 58 12 system system NOUN fis-4216 58 13 cracksn cracksn NOUN fis-4216 58 14 . . PUNCT fis-4216 59 1 the the DET fis-4216 59 2 proposed propose VERB fis-4216 59 3 system system NOUN fis-4216 59 4 consists consist VERB fis-4216 59 5 of of ADP fis-4216 59 6 three three NUM fis-4216 59 7 main main ADJ fis-4216 59 8 stages stage NOUN fis-4216 59 9 : : PUNCT fis-4216 59 10 augmentation augmentation NOUN fis-4216 59 11 of of ADP fis-4216 59 12 the the DET fis-4216 59 13 image image NOUN fis-4216 59 14 dataset dataset NOUN fis-4216 59 15 , , PUNCT fis-4216 59 16 training training NOUN fis-4216 59 17 and and CCONJ fis-4216 59 18 validation validation NOUN fis-4216 59 19 of of ADP fis-4216 59 20 the the DET fis-4216 59 21 adam adam NOUN fis-4216 59 22 - - PUNCT fis-4216 59 23 squeezenet squeezenet NOUN fis-4216 59 24 model model NOUN fis-4216 59 25 , , PUNCT fis-4216 59 26 and and CCONJ fis-4216 59 27 the the DET fis-4216 59 28 testing testing NOUN fis-4216 59 29 stage stage NOUN fis-4216 59 30 for for ADP fis-4216 59 31 decision decision NOUN fis-4216 59 32 - - PUNCT fis-4216 59 33 making making NOUN fis-4216 59 34 . . PUNCT fis-4216 60 1 this this DET fis-4216 60 2 system system NOUN fis-4216 60 3 aims aim VERB fis-4216 60 4 to to PART fis-4216 60 5 classify classify VERB fis-4216 60 6 the the DET fis-4216 60 7 digital digital ADJ fis-4216 60 8 image image NOUN fis-4216 60 9 data datum NOUN fis-4216 60 10 captured capture VERB fis-4216 60 11 by by ADP fis-4216 60 12 smartphones smartphone NOUN fis-4216 60 13 or or CCONJ fis-4216 60 14 uavs uavs NOUN fis-4216 60 15 . . PUNCT fis-4216 61 1 in in ADP fis-4216 61 2 the the DET fis-4216 61 3 first first ADJ fis-4216 61 4 stage stage NOUN fis-4216 61 5 , , PUNCT fis-4216 61 6 the the DET fis-4216 61 7 offline offline ADJ fis-4216 61 8 augmentation augmentation NOUN fis-4216 61 9 operation operation NOUN fis-4216 61 10 is be AUX fis-4216 61 11 performed perform VERB fis-4216 61 12 on on ADP fis-4216 61 13 the the DET fis-4216 61 14 captured capture VERB fis-4216 61 15 images image NOUN fis-4216 61 16 to to PART fis-4216 61 17 overcome overcome VERB fis-4216 61 18 the the DET fis-4216 61 19 possible possible ADJ fis-4216 61 20 imbalance imbalance NOUN fis-4216 61 21 problem problem NOUN fis-4216 61 22 . . PUNCT fis-4216 62 1 the the DET fis-4216 62 2 augmented augment VERB fis-4216 62 3 dataset dataset NOUN fis-4216 62 4 is be AUX fis-4216 62 5 split split VERB fis-4216 62 6 into into ADP fis-4216 62 7 the the DET fis-4216 62 8 train train NOUN fis-4216 62 9 , , PUNCT fis-4216 62 10 validation validation NOUN fis-4216 62 11 , , PUNCT fis-4216 62 12 and and CCONJ fis-4216 62 13 test test NOUN fis-4216 62 14 subsets subset NOUN fis-4216 62 15 without without ADP fis-4216 62 16 overlap overlap NOUN fis-4216 62 17 . . PUNCT fis-4216 63 1 the the DET fis-4216 63 2 training training NOUN fis-4216 63 3 dataset dataset NOUN fis-4216 63 4 comprises comprise VERB fis-4216 63 5 the the DET fis-4216 63 6 image image NOUN fis-4216 63 7 patch patch NOUN fis-4216 63 8 of of ADP fis-4216 63 9 the the DET fis-4216 63 10 concrete concrete ADJ fis-4216 63 11 surface surface NOUN fis-4216 63 12 and and CCONJ fis-4216 63 13 is be AUX fis-4216 63 14 set set VERB fis-4216 63 15 as as ADP fis-4216 63 16 the the DET fis-4216 63 17 input input NOUN fis-4216 63 18 of of ADP fis-4216 63 19 the the DET fis-4216 63 20 cracksn cracksn NOUN fis-4216 63 21 system system NOUN fis-4216 63 22 for for ADP fis-4216 63 23 parameter parameter NOUN fis-4216 63 24 optimization optimization NOUN fis-4216 63 25 of of ADP fis-4216 63 26 the the DET fis-4216 63 27 squeezenet squeezenet NOUN fis-4216 63 28 network network NOUN fis-4216 63 29 . . PUNCT fis-4216 64 1 the the DET fis-4216 64 2 validation validation NOUN fis-4216 64 3 set set NOUN fis-4216 64 4 contains contain VERB fis-4216 64 5 patches patch NOUN fis-4216 64 6 different different ADJ fis-4216 64 7 from from ADP fis-4216 64 8 the the DET fis-4216 64 9 train train NOUN fis-4216 64 10 set set NOUN fis-4216 64 11 to to PART fis-4216 64 12 evaluate evaluate VERB fis-4216 64 13 the the DET fis-4216 64 14 trained train VERB fis-4216 64 15 squeezenet squeezenet NOUN fis-4216 64 16 model model NOUN fis-4216 64 17 . . PUNCT fis-4216 65 1 the the DET fis-4216 65 2 hyperparameters hyperparameter NOUN fis-4216 65 3 of of ADP fis-4216 65 4 squeezenet squeezenet NOUN fis-4216 65 5 will will AUX fis-4216 65 6 be be AUX fis-4216 65 7 tuned tune VERB fis-4216 65 8 based base VERB fis-4216 65 9 on on ADP fis-4216 65 10 the the DET fis-4216 65 11 performance performance NOUN fis-4216 65 12 of of ADP fis-4216 65 13 validation validation NOUN fis-4216 65 14 . . PUNCT fis-4216 66 1 the the DET fis-4216 66 2 test test NOUN fis-4216 66 3 set set NOUN fis-4216 66 4 is be AUX fis-4216 66 5 taken take VERB fis-4216 66 6 as as ADP fis-4216 66 7 the the DET fis-4216 66 8 testing testing NOUN fis-4216 66 9 input input NOUN fis-4216 66 10 for for ADP fis-4216 66 11 the the DET fis-4216 66 12 optimized optimize VERB fis-4216 66 13 cracksn cracksn NOUN fis-4216 66 14 system system NOUN fis-4216 66 15 to to PART fis-4216 66 16 provide provide VERB fis-4216 66 17 an an DET fis-4216 66 18 unbiased unbiased ADJ fis-4216 66 19 final final ADJ fis-4216 66 20 evaluation evaluation NOUN fis-4216 66 21 of of ADP fis-4216 66 22 the the DET fis-4216 66 23 trained train VERB fis-4216 66 24 squeezenet squeezenet NOUN fis-4216 66 25 model model NOUN fis-4216 66 26 . . PUNCT fis-4216 67 1 the the DET fis-4216 67 2 squeezenet squeezenet NOUN fis-4216 67 3 convolution convolution NOUN fis-4216 67 4 network network NOUN fis-4216 67 5 is be AUX fis-4216 67 6 adopted adopt VERB fis-4216 67 7 in in ADP fis-4216 67 8 the the DET fis-4216 67 9 presented present VERB fis-4216 67 10 system system NOUN fis-4216 67 11 . . PUNCT fis-4216 68 1 it it PRON fis-4216 68 2 utilizes utilize VERB fis-4216 68 3 the the DET fis-4216 68 4 fire fire NOUN fis-4216 68 5 module module NOUN fis-4216 68 6 , , PUNCT fis-4216 68 7 incorporating incorporate VERB fis-4216 68 8 squeeze squeeze NOUN fis-4216 68 9 and and CCONJ fis-4216 68 10 expand expand VERB fis-4216 68 11 layers layer NOUN fis-4216 68 12 , , PUNCT fis-4216 68 13 so so ADV fis-4216 68 14 a a DET fis-4216 68 15 smaller small ADJ fis-4216 68 16 and and CCONJ fis-4216 68 17 more more ADV fis-4216 68 18 effective effective ADJ fis-4216 68 19 cnn cnn NOUN fis-4216 68 20 architecture architecture NOUN fis-4216 68 21 is be AUX fis-4216 68 22 constructed construct VERB fis-4216 68 23 . . PUNCT fis-4216 69 1 adams adams PROPN fis-4216 69 2 optimization optimization PROPN fis-4216 69 3 is be AUX fis-4216 69 4 utilized utilize VERB fis-4216 69 5 in in ADP fis-4216 69 6 the the DET fis-4216 69 7 cnn cnn PROPN fis-4216 69 8 network network NOUN fis-4216 69 9 in in ADP fis-4216 69 10 order order NOUN fis-4216 69 11 to to PART fis-4216 69 12 obtain obtain VERB fis-4216 69 13 the the DET fis-4216 69 14 best good ADJ fis-4216 69 15 decision decision NOUN fis-4216 69 16 - - PUNCT fis-4216 69 17 making make VERB fis-4216 69 18 model model NOUN fis-4216 69 19 . . PUNCT fis-4216 70 1 the the DET fis-4216 70 2 best good ADJ fis-4216 70 3 squeezenet squeezenet NOUN fis-4216 70 4 model model NOUN fis-4216 70 5 is be AUX fis-4216 70 6 later later ADV fis-4216 70 7 used use VERB fis-4216 70 8 for for ADP fis-4216 70 9 the the DET fis-4216 70 10 decision decision NOUN fis-4216 70 11 - - PUNCT fis-4216 70 12 making make VERB fis-4216 70 13 process process NOUN fis-4216 70 14 with with ADP fis-4216 70 15 the the DET fis-4216 70 16 test test NOUN fis-4216 70 17 set set NOUN fis-4216 70 18 . . PUNCT fis-4216 71 1 the the DET fis-4216 71 2 optimized optimize VERB fis-4216 71 3 network network NOUN fis-4216 71 4 model model NOUN fis-4216 71 5 classifies classify VERB fis-4216 71 6 the the DET fis-4216 71 7 images image NOUN fis-4216 71 8 in in ADP fis-4216 71 9 the the DET fis-4216 71 10 test test NOUN fis-4216 71 11 dataset dataset NOUN fis-4216 71 12 , , PUNCT fis-4216 71 13 and and CCONJ fis-4216 71 14 the the DET fis-4216 71 15 classification classification NOUN fis-4216 71 16 performances performance NOUN fis-4216 71 17 are be AUX fis-4216 71 18 determined determine VERB fis-4216 71 19 . . PUNCT fis-4216 72 1 figure figure VERB fis-4216 72 2 1 1 NUM fis-4216 72 3 : : PUNCT fis-4216 72 4 architecture architecture NOUN fis-4216 72 5 of of ADP fis-4216 72 6 the the DET fis-4216 72 7 deep deep ADJ fis-4216 72 8 learning learning NOUN fis-4216 72 9 crack crack VERB fis-4216 72 10 detection detection NOUN fis-4216 72 11 system system NOUN fis-4216 72 12 . . PUNCT fis-4216 73 1 image image NOUN fis-4216 73 2 data datum NOUN fis-4216 73 3 preparation preparation NOUN fis-4216 73 4 an an DET fis-4216 73 5 image image NOUN fis-4216 73 6 dataset dataset NOUN fis-4216 73 7 of of ADP fis-4216 73 8 concrete concrete ADJ fis-4216 73 9 crack crack NOUN fis-4216 73 10 images image NOUN fis-4216 73 11 for for ADP fis-4216 73 12 classification classification NOUN fis-4216 73 13 was be AUX fis-4216 73 14 obtained obtain VERB fis-4216 73 15 from from ADP fis-4216 73 16 a a DET fis-4216 73 17 total total NOUN fis-4216 73 18 of of ADP fis-4216 73 19 284 284 NUM fis-4216 73 20 images image NOUN fis-4216 73 21 ( ( PUNCT fis-4216 73 22 1706×1280 1706×1280 NUM fis-4216 73 23 pixels pixel NOUN fis-4216 73 24 , , PUNCT fis-4216 73 25 8 8 NUM fis-4216 73 26 bits bit NOUN fis-4216 73 27 , , PUNCT fis-4216 73 28 rgb rgb PROPN fis-4216 73 29 channels channel NOUN fis-4216 73 30 ) ) PUNCT fis-4216 73 31 taken take VERB fis-4216 73 32 on on ADP fis-4216 73 33 the the DET fis-4216 73 34 campus campus NOUN fis-4216 73 35 of of ADP fis-4216 73 36 hefei hefei PROPN fis-4216 73 37 university university PROPN fis-4216 73 38 by by ADP fis-4216 73 39 a a DET fis-4216 73 40 smartphone smartphone NOUN fis-4216 73 41 . . PUNCT fis-4216 74 1 following follow VERB fis-4216 74 2 the the DET fis-4216 74 3 same same ADJ fis-4216 74 4 methodology methodology NOUN fis-4216 74 5 outlined outline VERB fis-4216 74 6 in in ADP fis-4216 74 7 [ [ X fis-4216 74 8 11 11 NUM fis-4216 74 9 ] ] PUNCT fis-4216 74 10 , , PUNCT fis-4216 74 11 the the DET fis-4216 74 12 full full ADJ fis-4216 74 13 - - PUNCT fis-4216 74 14 size size NOUN fis-4216 74 15 , , PUNCT fis-4216 74 16 high high ADJ fis-4216 74 17 - - PUNCT fis-4216 74 18 resolution resolution NOUN fis-4216 74 19 images image NOUN fis-4216 74 20 were be AUX fis-4216 74 21 cropped crop VERB fis-4216 74 22 , , PUNCT fis-4216 74 23 resized resize VERB fis-4216 74 24 into into ADP fis-4216 74 25 spatially spatially ADV fis-4216 74 26 sized sized ADJ fis-4216 74 27 square square ADJ fis-4216 74 28 patches patch NOUN fis-4216 74 29 , , PUNCT fis-4216 74 30 and and CCONJ fis-4216 74 31 labeled label VERB fis-4216 74 32 into into ADP fis-4216 74 33 two two NUM fis-4216 74 34 classes class NOUN fis-4216 74 35 . . PUNCT fis-4216 75 1 a a PRON fis-4216 75 2 sub sub NOUN fis-4216 75 3 - - ADJ fis-4216 75 4 image image NOUN fis-4216 75 5 was be AUX fis-4216 75 6 labeled label VERB fis-4216 75 7 as as ADP fis-4216 75 8 a a DET fis-4216 75 9 cracked cracked ADJ fis-4216 75 10 ( ( PUNCT fis-4216 75 11 i.e. i.e. X fis-4216 75 12 , , PUNCT fis-4216 75 13 positive positive ADJ fis-4216 75 14 ) ) PUNCT fis-4216 75 15 patch patch NOUN fis-4216 75 16 if if SCONJ fis-4216 75 17 it it PRON fis-4216 75 18 was be AUX fis-4216 75 19 centered center VERB fis-4216 75 20 10 10 NUM fis-4216 75 21 pixels pixel NOUN fis-4216 75 22 from from ADP fis-4216 75 23 the the DET fis-4216 75 24 crack crack NOUN fis-4216 75 25 centroid centroid NOUN fis-4216 75 26 ; ; PUNCT fis-4216 75 27 otherwise otherwise ADV fis-4216 75 28 , , PUNCT fis-4216 75 29 it it PRON fis-4216 75 30 was be AUX fis-4216 75 31 classified classify VERB fis-4216 75 32 as as ADP fis-4216 75 33 a a DET fis-4216 75 34 normal normal ADJ fis-4216 75 35 ( ( PUNCT fis-4216 75 36 i.e. i.e. X fis-4216 75 37 , , PUNCT fis-4216 75 38 negative negative ADJ fis-4216 75 39 ) ) PUNCT fis-4216 75 40 patch patch NOUN fis-4216 75 41 . . PUNCT fis-4216 76 1 to to PART fis-4216 76 2 reduce reduce VERB fis-4216 76 3 the the DET fis-4216 76 4 patch patch ADJ fis-4216 76 5 similarity similarity NOUN fis-4216 76 6 in in ADP fis-4216 76 7 the the DET fis-4216 76 8 image image NOUN fis-4216 76 9 dataset dataset NOUN fis-4216 76 10 , , PUNCT fis-4216 76 11 the the DET fis-4216 76 12 overlap overlap NOUN fis-4216 76 13 of of ADP fis-4216 76 14 two two NUM fis-4216 76 15 cracked crack VERB fis-4216 76 16 or or CCONJ fis-4216 76 17 positive positive ADJ fis-4216 76 18 patches patch NOUN fis-4216 76 19 p(1 p(1 PROPN fis-4216 76 20 ) ) PUNCT fis-4216 76 21 and and CCONJ fis-4216 76 22 p(2 p(2 NOUN fis-4216 76 23 ) ) PUNCT fis-4216 76 24 is be AUX fis-4216 76 25 defined define VERB fis-4216 76 26 as as ADP fis-4216 76 27 o o NOUN fis-4216 76 28 = = NOUN fis-4216 76 29 area(p(1 area(p(1 ADJ fis-4216 76 30 ) ) PUNCT fis-4216 76 31 ∩ ∩ NOUN fis-4216 76 32 p(2))/area(p(1)∪p(2 p(2))/area(p(1)∪p(2 X fis-4216 76 33 ) ) PUNCT fis-4216 76 34 ) ) PUNCT fis-4216 76 35 should should AUX fis-4216 76 36 be be AUX fis-4216 76 37 maintained maintain VERB fis-4216 76 38 at at ADP fis-4216 76 39 a a DET fis-4216 76 40 low low ADJ fis-4216 76 41 level level NOUN fis-4216 76 42 . . PUNCT fis-4216 77 1 the the DET fis-4216 77 2 distance distance NOUN fis-4216 77 3 between between ADP fis-4216 77 4 the the DET fis-4216 77 5 centers center NOUN fis-4216 77 6 of of ADP fis-4216 77 7 two two NUM fis-4216 77 8 adjacent adjacent ADJ fis-4216 77 9 patches patch NOUN fis-4216 77 10 was be AUX fis-4216 77 11 set set VERB fis-4216 77 12 to to ADP fis-4216 77 13 0.7 0.7 NUM fis-4216 77 14 times time NOUN fis-4216 77 15 the the DET fis-4216 77 16 patch patch ADJ fis-4216 77 17 width width NOUN fis-4216 77 18 . . PUNCT fis-4216 78 1 the the DET fis-4216 78 2 patches patch NOUN fis-4216 78 3 were be AUX fis-4216 78 4 resized resize VERB fis-4216 78 5 to to ADP fis-4216 78 6 a a DET fis-4216 78 7 fixed fix VERB fis-4216 78 8 size size NOUN fis-4216 78 9 of of ADP fis-4216 78 10 227×227 227×227 NUM fis-4216 78 11 pixels pixel NOUN fis-4216 78 12 to to PART fis-4216 78 13 accommodate accommodate VERB fis-4216 78 14 the the DET fis-4216 78 15 required require VERB fis-4216 78 16 input input NOUN fis-4216 78 17 image image NOUN fis-4216 78 18 size size NOUN fis-4216 78 19 of of ADP fis-4216 78 20 the the DET fis-4216 78 21 squeezenet squeezenet NOUN fis-4216 78 22 model model NOUN fis-4216 78 23 . . PUNCT fis-4216 79 1 since since SCONJ fis-4216 79 2 crack crack NOUN fis-4216 79 3 features feature NOUN fis-4216 79 4 comprise comprise VERB fis-4216 79 5 only only ADV fis-4216 79 6 a a DET fis-4216 79 7 small small ADJ fis-4216 79 8 portion portion NOUN fis-4216 79 9 of of ADP fis-4216 79 10 the the DET fis-4216 79 11 collected collect VERB fis-4216 79 12 images image NOUN fis-4216 79 13 , , PUNCT fis-4216 79 14 the the DET fis-4216 79 15 sub sub NOUN fis-4216 79 16 - - NOUN fis-4216 79 17 image image NOUN fis-4216 79 18 was be AUX fis-4216 79 19 rotated rotate VERB fis-4216 79 20 around around ADP fis-4216 79 21 its its PRON fis-4216 79 22 center center NOUN fis-4216 79 23 by by ADP fis-4216 79 24 a a DET fis-4216 79 25 random random ADJ fis-4216 79 26 angle angle NOUN fis-4216 79 27 γ∈ γ∈ X fis-4216 80 1 [ [ X fis-4216 80 2 0 0 NUM fis-4216 80 3 , , PUNCT fis-4216 80 4 360 360 NUM fis-4216 80 5 ] ] PUNCT fis-4216 80 6 to to PART fis-4216 80 7 boost boost VERB fis-4216 80 8 the the DET fis-4216 80 9 total total ADJ fis-4216 80 10 number number NOUN fis-4216 80 11 of of ADP fis-4216 80 12 cracked crack VERB fis-4216 80 13 patches patch NOUN fis-4216 80 14 . . PUNCT fis-4216 81 1 the the DET fis-4216 81 2 inadequacy inadequacy NOUN fis-4216 81 3 of of ADP fis-4216 81 4 the the DET fis-4216 81 5 image image NOUN fis-4216 81 6 data datum NOUN fis-4216 81 7 has have VERB fis-4216 81 8 a a DET fis-4216 81 9 detrimental detrimental ADJ fis-4216 81 10 effect effect NOUN fis-4216 81 11 on on ADP fis-4216 81 12 classification classification NOUN fis-4216 81 13 performance performance NOUN fis-4216 81 14 [ [ X fis-4216 81 15 19 19 NUM fis-4216 81 16 ] ] PUNCT fis-4216 81 17 . . PUNCT fis-4216 82 1 in in ADP fis-4216 82 2 order order NOUN fis-4216 82 3 to to PART fis-4216 82 4 overcome overcome VERB fis-4216 82 5 the the DET fis-4216 82 6 imbalance imbalance NOUN fis-4216 82 7 problem problem NOUN fis-4216 82 8 of of ADP fis-4216 82 9 the the DET fis-4216 82 10 image image NOUN fis-4216 82 11 dataset dataset NOUN fis-4216 82 12 , , PUNCT fis-4216 82 13 the the DET fis-4216 82 14 sub sub NOUN fis-4216 82 15 - - NOUN fis-4216 82 16 images image NOUN fis-4216 82 17 were be AUX fis-4216 82 18 augmented augment VERB fis-4216 82 19 by by ADP fis-4216 82 20 flipping flip VERB fis-4216 82 21 them they PRON fis-4216 82 22 along along ADP fis-4216 82 23 their their PRON fis-4216 82 24 vertical vertical ADJ fis-4216 82 25 axis axis NOUN fis-4216 82 26 and and CCONJ fis-4216 82 27 randomly randomly ADV fis-4216 82 28 shifting shift VERB fis-4216 82 29 them they PRON fis-4216 82 30 up up ADP fis-4216 82 31 to to PART fis-4216 82 32 30 30 NUM fis-4216 82 33 pixels pixel NOUN fis-4216 82 34 in in ADP fis-4216 82 35 either either DET fis-4216 82 36 direction direction NOUN fis-4216 82 37 or or CCONJ fis-4216 82 38 both both DET fis-4216 82 39 directions direction NOUN fis-4216 82 40 . . PUNCT fis-4216 83 1 the the DET fis-4216 83 2 augmentation augmentation NOUN fis-4216 83 3 operation operation NOUN fis-4216 83 4 also also ADV fis-4216 83 5 prevents prevent VERB fis-4216 83 6 the the DET fis-4216 83 7 network network NOUN fis-4216 83 8 from from ADP fis-4216 83 9 overfitting overfitte VERB fis-4216 83 10 and and CCONJ fis-4216 83 11 memorizing memorize VERB fis-4216 83 12 the the DET fis-4216 83 13 exact exact ADJ fis-4216 83 14 details detail NOUN fis-4216 83 15 of of ADP fis-4216 83 16 the the DET fis-4216 83 17 image image NOUN fis-4216 83 18 data datum NOUN fis-4216 83 19 during during ADP fis-4216 83 20 the the DET fis-4216 83 21 training training NOUN fis-4216 83 22 process process NOUN fis-4216 83 23 . . PUNCT fis-4216 84 1 fig fig NOUN fis-4216 84 2 . . PUNCT fis-4216 85 1 2 2 NUM fis-4216 85 2 shows show VERB fis-4216 85 3 the the DET fis-4216 85 4 sample sample NOUN fis-4216 85 5 patches patch NOUN fis-4216 85 6 of of ADP fis-4216 85 7 the the DET fis-4216 85 8 concrete concrete ADJ fis-4216 85 9 surface surface NOUN fis-4216 85 10 with with ADP fis-4216 85 11 labels label NOUN fis-4216 85 12 , , PUNCT fis-4216 85 13 which which PRON fis-4216 85 14 will will AUX fis-4216 85 15 be be AUX fis-4216 85 16 used use VERB fis-4216 85 17 for for ADP fis-4216 85 18 network network NOUN fis-4216 85 19 training training NOUN fis-4216 85 20 and and CCONJ fis-4216 85 21 validation validation NOUN fis-4216 85 22 . . PUNCT fis-4216 86 1 the the DET fis-4216 86 2 obtained obtain VERB fis-4216 86 3 patches patch NOUN fis-4216 86 4 of of ADP fis-4216 86 5 concrete concrete ADJ fis-4216 86 6 surfaces surface NOUN fis-4216 86 7 demonstrate demonstrate VERB fis-4216 86 8 variety variety NOUN fis-4216 86 9 in in ADP fis-4216 86 10 terms term NOUN fis-4216 86 11 of of ADP fis-4216 86 12 texture texture NOUN fis-4216 86 13 , , PUNCT fis-4216 86 14 surface surface NOUN fis-4216 86 15 finish finish NOUN fis-4216 86 16 , , PUNCT fis-4216 86 17 illumination illumination NOUN fis-4216 86 18 , , PUNCT fis-4216 86 19 and and CCONJ fis-4216 86 20 focus focus VERB fis-4216 86 21 . . PUNCT fis-4216 87 1 the the DET fis-4216 87 2 obtained obtain VERB fis-4216 87 3 dataset dataset NOUN fis-4216 87 4 contains contain VERB fis-4216 87 5 6,000 6,000 NUM fis-4216 87 6 images image NOUN fis-4216 87 7 , , PUNCT fis-4216 87 8 half half NOUN fis-4216 87 9 of of ADP fis-4216 87 10 which which PRON fis-4216 87 11 are be AUX fis-4216 87 12 cracked crack VERB fis-4216 87 13 or or CCONJ fis-4216 87 14 positive positive ADJ fis-4216 87 15 patches patch NOUN fis-4216 87 16 . . PUNCT fis-4216 88 1 by by ADP fis-4216 88 2 random random ADJ fis-4216 88 3 selection selection NOUN fis-4216 88 4 , , PUNCT fis-4216 88 5 the the DET fis-4216 88 6 images image NOUN fis-4216 88 7 after after SCONJ fis-4216 88 8 offline offline ADJ fis-4216 88 9 augmentation augmentation NOUN fis-4216 88 10 was be AUX fis-4216 88 11 split split VERB fis-4216 88 12 into into ADP fis-4216 88 13 the the DET fis-4216 88 14 train train NOUN fis-4216 88 15 , , PUNCT fis-4216 88 16 validation validation NOUN fis-4216 88 17 , , PUNCT fis-4216 88 18 and and CCONJ fis-4216 88 19 test test NOUN fis-4216 88 20 image image NOUN fis-4216 88 21 datasets dataset NOUN fis-4216 88 22 . . PUNCT fis-4216 89 1 the the DET fis-4216 89 2 number number NOUN fis-4216 89 3 of of ADP fis-4216 89 4 images image NOUN fis-4216 89 5 in in ADP fis-4216 89 6 the the DET fis-4216 89 7 three three NUM fis-4216 89 8 corresponding correspond VERB fis-4216 89 9 datasets dataset NOUN fis-4216 89 10 is be AUX fis-4216 89 11 4000 4000 NUM fis-4216 89 12 , , PUNCT fis-4216 89 13 1000 1000 NUM fis-4216 89 14 , , PUNCT fis-4216 89 15 and and CCONJ fis-4216 89 16 1000 1000 NUM fis-4216 89 17 , , PUNCT fis-4216 89 18 respectively respectively ADV fis-4216 89 19 . . PUNCT fis-4216 90 1 l. l. PROPN fis-4216 90 2 wang wang PROPN fis-4216 90 3 , , PUNCT fis-4216 90 4 frattura frattura PROPN fis-4216 90 5 ed ed PROPN fis-4216 90 6 integrità integrità PROPN fis-4216 90 7 strutturale strutturale PROPN fis-4216 90 8 , , PUNCT fis-4216 90 9 65 65 NUM fis-4216 90 10 ( ( PUNCT fis-4216 90 11 2023 2023 NUM fis-4216 90 12 ) ) PUNCT fis-4216 90 13 289 289 NUM fis-4216 90 14 - - SYM fis-4216 90 15 299 299 NUM fis-4216 90 16 ; ; PUNCT fis-4216 90 17 doi doi NOUN fis-4216 90 18 : : PUNCT fis-4216 90 19 10.3221 10.3221 NUM fis-4216 90 20 / / SYM fis-4216 90 21 igf igf PROPN fis-4216 90 22 - - PUNCT fis-4216 90 23 esis.65.19 esis.65.19 NOUN fis-4216 90 24 292 292 NUM fis-4216 90 25 given give VERB fis-4216 90 26 the the DET fis-4216 90 27 selection selection NOUN fis-4216 90 28 's 's PART fis-4216 90 29 randomness randomness NOUN fis-4216 90 30 and and CCONJ fis-4216 90 31 the the DET fis-4216 90 32 data datum NOUN fis-4216 90 33 's 's PART fis-4216 90 34 abundance abundance NOUN fis-4216 90 35 , , PUNCT fis-4216 90 36 the the DET fis-4216 90 37 normal normal ADJ fis-4216 90 38 and and CCONJ fis-4216 90 39 cracked crack VERB fis-4216 90 40 images image NOUN fis-4216 90 41 are be AUX fis-4216 90 42 split split VERB fis-4216 90 43 almost almost ADV fis-4216 90 44 equally equally ADV fis-4216 90 45 among among ADP fis-4216 90 46 the the DET fis-4216 90 47 three three NUM fis-4216 90 48 image image NOUN fis-4216 90 49 datasets dataset NOUN fis-4216 90 50 . . PUNCT fis-4216 91 1 figure figure NOUN fis-4216 91 2 2 2 NUM fis-4216 91 3 : : PUNCT fis-4216 91 4 concrete concrete ADJ fis-4216 91 5 image image NOUN fis-4216 91 6 with with ADP fis-4216 91 7 cracks crack NOUN fis-4216 91 8 and and CCONJ fis-4216 91 9 image image NOUN fis-4216 91 10 patches patch NOUN fis-4216 91 11 obtained obtain VERB fis-4216 91 12 through through ADP fis-4216 91 13 cropping cropping NOUN fis-4216 91 14 , , PUNCT fis-4216 91 15 flipping flip VERB fis-4216 91 16 , , PUNCT fis-4216 91 17 and and CCONJ fis-4216 91 18 rotating rotate VERB fis-4216 91 19 augmentation augmentation NOUN fis-4216 91 20 . . PUNCT fis-4216 92 1 squeezenet squeezenet NOUN fis-4216 92 2 architecture architecture NOUN fis-4216 92 3 a a DET fis-4216 92 4 deep deep ADJ fis-4216 92 5 learning learning NOUN fis-4216 92 6 algorithm algorithm NOUN fis-4216 92 7 based base VERB fis-4216 92 8 on on ADP fis-4216 92 9 squeezenet squeezenet NOUN fis-4216 92 10 cracksn cracksn NOUN fis-4216 92 11 is be AUX fis-4216 92 12 implemented implement VERB fis-4216 92 13 for for ADP fis-4216 92 14 automatic automatic ADJ fis-4216 92 15 crack crack NOUN fis-4216 92 16 detection detection NOUN fis-4216 92 17 . . PUNCT fis-4216 93 1 the the DET fis-4216 93 2 squeezenet squeezenet NOUN fis-4216 93 3 is be AUX fis-4216 93 4 a a DET fis-4216 93 5 cnn cnn PROPN fis-4216 93 6 - - PUNCT fis-4216 93 7 based base VERB fis-4216 93 8 architecture architecture NOUN fis-4216 93 9 with with ADP fis-4216 93 10 only only ADV fis-4216 93 11 1/50 1/50 NUM fis-4216 93 12 parameters parameter NOUN fis-4216 93 13 while while SCONJ fis-4216 93 14 maintaining maintain VERB fis-4216 93 15 the the DET fis-4216 93 16 competitive competitive ADJ fis-4216 93 17 accuracy accuracy NOUN fis-4216 93 18 of of ADP fis-4216 93 19 the the DET fis-4216 93 20 alexnet alexnet NOUN fis-4216 94 1 [ [ X fis-4216 94 2 20 20 NUM fis-4216 94 3 ] ] PUNCT fis-4216 94 4 . . PUNCT fis-4216 95 1 in in ADP fis-4216 95 2 contrast contrast NOUN fis-4216 95 3 to to ADP fis-4216 95 4 cnns cnns PROPN fis-4216 95 5 , , PUNCT fis-4216 95 6 squeezenet squeezenet NOUN fis-4216 95 7 replaces replace VERB fis-4216 95 8 the the DET fis-4216 95 9 filter filter NOUN fis-4216 95 10 or or CCONJ fis-4216 95 11 kernel kernel NOUN fis-4216 95 12 size size NOUN fis-4216 95 13 from from ADP fis-4216 95 14 3×3 3×3 NUM fis-4216 95 15 to to ADP fis-4216 95 16 1×1 1×1 NUM fis-4216 95 17 and and CCONJ fis-4216 95 18 reduces reduce VERB fis-4216 95 19 the the DET fis-4216 95 20 number number NOUN fis-4216 95 21 of of ADP fis-4216 95 22 input input NOUN fis-4216 95 23 channels channel NOUN fis-4216 95 24 to to ADP fis-4216 95 25 3×3 3×3 NUM fis-4216 95 26 filters filter NOUN fis-4216 95 27 . . PUNCT fis-4216 96 1 these these DET fis-4216 96 2 strategies strategy NOUN fis-4216 96 3 bring bring VERB fis-4216 96 4 significant significant ADJ fis-4216 96 5 parameter parameter NOUN fis-4216 96 6 size size NOUN fis-4216 96 7 reductions reduction NOUN fis-4216 96 8 to to PART fis-4216 96 9 cnns cnns VERB fis-4216 96 10 while while SCONJ fis-4216 96 11 maintaining maintain VERB fis-4216 96 12 the the DET fis-4216 96 13 prediction prediction NOUN fis-4216 96 14 accuracy accuracy NOUN fis-4216 96 15 of of ADP fis-4216 96 16 the the DET fis-4216 96 17 algorithm algorithm NOUN fis-4216 96 18 . . PUNCT fis-4216 97 1 to to PART fis-4216 97 2 optimize optimize VERB fis-4216 97 3 its its PRON fis-4216 97 4 accuracy accuracy NOUN fis-4216 97 5 with with ADP fis-4216 97 6 a a DET fis-4216 97 7 limited limited ADJ fis-4216 97 8 parameter parameter NOUN fis-4216 97 9 size size NOUN fis-4216 97 10 , , PUNCT fis-4216 97 11 squeezenet squeezenet NOUN fis-4216 97 12 executes execute VERB fis-4216 97 13 down down ADV fis-4216 97 14 - - PUNCT fis-4216 97 15 sampling sampling NOUN fis-4216 97 16 later later ADV fis-4216 97 17 in in ADP fis-4216 97 18 the the DET fis-4216 97 19 network network NOUN fis-4216 97 20 so so SCONJ fis-4216 97 21 that that SCONJ fis-4216 97 22 convolutional convolutional ADJ fis-4216 97 23 layers layer NOUN fis-4216 97 24 have have VERB fis-4216 97 25 large large ADJ fis-4216 97 26 activation activation NOUN fis-4216 97 27 maps map NOUN fis-4216 97 28 . . PUNCT fis-4216 98 1 as as SCONJ fis-4216 98 2 illustrated illustrate VERB fis-4216 98 3 in in ADP fis-4216 98 4 fig fig NOUN fis-4216 98 5 . . PUNCT fis-4216 99 1 1 1 NUM fis-4216 99 2 , , PUNCT fis-4216 99 3 the the DET fis-4216 99 4 augmented augment VERB fis-4216 99 5 image image NOUN fis-4216 99 6 datasets dataset NOUN fis-4216 99 7 were be AUX fis-4216 99 8 used use VERB fis-4216 99 9 as as ADP fis-4216 99 10 input input NOUN fis-4216 99 11 for for ADP fis-4216 99 12 network network NOUN fis-4216 99 13 training training NOUN fis-4216 99 14 and and CCONJ fis-4216 99 15 optimization optimization NOUN fis-4216 99 16 . . PUNCT fis-4216 100 1 the the DET fis-4216 100 2 squeezenet squeezenet NOUN fis-4216 100 3 starts start VERB fis-4216 100 4 with with ADP fis-4216 100 5 a a DET fis-4216 100 6 standalone standalone ADJ fis-4216 100 7 convolution convolution NOUN fis-4216 100 8 layer layer NOUN fis-4216 100 9 ( ( PUNCT fis-4216 100 10 conv1 conv1 PROPN fis-4216 100 11 ) ) PUNCT fis-4216 100 12 , , PUNCT fis-4216 100 13 followed follow VERB fis-4216 100 14 by by ADP fis-4216 100 15 eight eight NUM fis-4216 100 16 fire fire NOUN fis-4216 100 17 modules module NOUN fis-4216 100 18 ( ( PUNCT fis-4216 100 19 fire2 fire2 PROPN fis-4216 100 20 to to PART fis-4216 100 21 fire9 fire9 PROPN fis-4216 100 22 ) ) PUNCT fis-4216 100 23 , , PUNCT fis-4216 100 24 and and CCONJ fis-4216 100 25 ends end VERB fis-4216 100 26 with with ADP fis-4216 100 27 a a DET fis-4216 100 28 final final ADJ fis-4216 100 29 convolution convolution NOUN fis-4216 100 30 layer layer NOUN fis-4216 100 31 ( ( PUNCT fis-4216 100 32 conv10 conv10 PROPN fis-4216 100 33 ) ) PUNCT fis-4216 100 34 . . PUNCT fis-4216 101 1 the the DET fis-4216 101 2 entire entire ADJ fis-4216 101 3 layer layer NOUN fis-4216 101 4 configuration configuration NOUN fis-4216 101 5 of of ADP fis-4216 101 6 the the DET fis-4216 101 7 squeezenet squeezenet NOUN fis-4216 101 8 - - PUNCT fis-4216 101 9 based base VERB fis-4216 101 10 architecture architecture NOUN fis-4216 101 11 is be AUX fis-4216 101 12 presented present VERB fis-4216 101 13 in in ADP fis-4216 101 14 tab tab NOUN fis-4216 101 15 . . PUNCT fis-4216 102 1 1 1 X fis-4216 102 2 . . X fis-4216 102 3 a a DET fis-4216 102 4 convolution convolution NOUN fis-4216 102 5 layer layer NOUN fis-4216 102 6 is be AUX fis-4216 102 7 used use VERB fis-4216 102 8 to to PART fis-4216 102 9 transform transform VERB fis-4216 102 10 an an DET fis-4216 102 11 input input NOUN fis-4216 102 12 image image NOUN fis-4216 102 13 into into ADP fis-4216 102 14 a a DET fis-4216 102 15 feature feature NOUN fis-4216 102 16 map map NOUN fis-4216 102 17 , , PUNCT fis-4216 102 18 as as SCONJ fis-4216 102 19 demonstrated demonstrate VERB fis-4216 102 20 in in ADP fis-4216 102 21 fig fig NOUN fis-4216 102 22 . . PUNCT fis-4216 103 1 3 3 X fis-4216 103 2 . . X fis-4216 103 3 the the DET fis-4216 103 4 filter filter NOUN fis-4216 103 5 converts convert VERB fis-4216 103 6 the the DET fis-4216 103 7 input input NOUN fis-4216 103 8 image image NOUN fis-4216 103 9 x x PUNCT fis-4216 103 10 into into ADP fis-4216 103 11 a a DET fis-4216 103 12 subsequent subsequent ADJ fis-4216 103 13 filtered filter VERB fis-4216 103 14 image image NOUN fis-4216 103 15 y y PROPN fis-4216 103 16 or or CCONJ fis-4216 103 17 feature feature NOUN fis-4216 103 18 map map NOUN fis-4216 103 19 , , PUNCT fis-4216 103 20 by by ADP fis-4216 103 21 ∑∑ > X fis-4216 103 25 i i PROPN fis-4216 103 26 , , PUNCT fis-4216 103 27 j j PROPN fis-4216 103 28 t t PROPN fis-4216 103 29 , , PUNCT fis-4216 103 30 r r PROPN fis-4216 103 31 i i PROPN fis-4216 103 32 - - PUNCT fis-4216 103 33 t t PROPN fis-4216 103 34 , , PUNCT fis-4216 103 35 j j PROPN fis-4216 103 36 - - PUNCT fis-4216 103 37 r r NOUN fis-4216 103 38 t t NOUN fis-4216 103 39 r r NOUN fis-4216 103 40 y y PROPN fis-4216 103 41 =( =( NOUN fis-4216 103 42 x x NOUN fis-4216 103 43     SPACE fis-4216 103 44 * * ADJ fis-4216 103 45 w w PROPN fis-4216 103 46 ) ) PUNCT fis-4216 103 47 = = SYM fis-4216 104 1 w w PROPN fis-4216 104 2 x x PUNCT fis-4216 104 3 with with ADP fis-4216 104 4 t t PROPN fis-4216 104 5 , , PUNCT fis-4216 104 6 r r NOUN fis-4216 104 7 = = NOUN fis-4216 104 8 0,1,2 0,1,2 NOUN fis-4216 104 9 . . PUNCT fis-4216 105 1 ( ( PUNCT fis-4216 105 2 1 1 X fis-4216 105 3 ) ) PUNCT fis-4216 105 4 the the DET fis-4216 105 5 output output NOUN fis-4216 105 6 dimension dimension NOUN fis-4216 105 7 of of ADP fis-4216 105 8 a a DET fis-4216 105 9 convolution convolution NOUN fis-4216 105 10 operation operation NOUN fis-4216 105 11 is be AUX fis-4216 105 12 determined determine VERB fis-4216 105 13 by by ADP fis-4216 105 14 two two NUM fis-4216 105 15 parameters parameter NOUN fis-4216 105 16 , , PUNCT fis-4216 105 17 namely namely ADV fis-4216 105 18 the the DET fis-4216 105 19 step step NOUN fis-4216 105 20 size size NOUN fis-4216 105 21 of of ADP fis-4216 105 22 stride stride NOUN fis-4216 105 23 s s NOUN fis-4216 105 24 and and CCONJ fis-4216 105 25 the the DET fis-4216 105 26 padding padding NOUN fis-4216 105 27 length length NOUN fis-4216 105 28 f. f. PROPN fis-4216 106 1 the the DET fis-4216 106 2 squeezenet squeezenet NOUN fis-4216 106 3 model model NOUN fis-4216 106 4 uniquely uniquely ADV fis-4216 106 5 defines define VERB fis-4216 106 6 the the DET fis-4216 106 7 fire fire NOUN fis-4216 106 8 module module NOUN fis-4216 106 9 as as ADP fis-4216 106 10 a a DET fis-4216 106 11 squeeze squeeze NOUN fis-4216 106 12 convolution convolution NOUN fis-4216 106 13 layer layer NOUN fis-4216 106 14 ( ( PUNCT fis-4216 106 15 1×1 1×1 NUM fis-4216 106 16 filter filter NOUN fis-4216 106 17 ) ) PUNCT fis-4216 106 18 that that PRON fis-4216 106 19 feeds feed VERB fis-4216 106 20 into into ADP fis-4216 106 21 an an DET fis-4216 106 22 expanded expand VERB fis-4216 106 23 layer layer NOUN fis-4216 106 24 with with ADP fis-4216 106 25 a a DET fis-4216 106 26 mix mix NOUN fis-4216 106 27 of of ADP fis-4216 106 28 1×1 1×1 NUM fis-4216 106 29 and and CCONJ fis-4216 106 30 3×3 3×3 NUM fis-4216 106 31 convolution convolution NOUN fis-4216 106 32 filters filter NOUN fis-4216 106 33 . . PUNCT fis-4216 107 1 both both CCONJ fis-4216 107 2 the the DET fis-4216 107 3 squeeze squeeze NOUN fis-4216 107 4 convolution convolution NOUN fis-4216 107 5 and and CCONJ fis-4216 107 6 expanded expand VERB fis-4216 107 7 layers layer NOUN fis-4216 107 8 are be AUX fis-4216 107 9 connected connect VERB fis-4216 107 10 to to ADP fis-4216 107 11 the the DET fis-4216 107 12 rectified rectify VERB fis-4216 107 13 linear linear PROPN fis-4216 107 14 unit unit NOUN fis-4216 107 15 relu relu NOUN fis-4216 107 16 activation activation NOUN fis-4216 107 17 function function NOUN fis-4216 107 18 . . PUNCT fis-4216 108 1 compared compare VERB fis-4216 108 2 to to ADP fis-4216 108 3 other other ADJ fis-4216 108 4 functions function NOUN fis-4216 108 5 in in ADP fis-4216 108 6 traditional traditional ADJ fis-4216 108 7 neural neural ADJ fis-4216 108 8 models model NOUN fis-4216 108 9 , , PUNCT fis-4216 108 10 relu relu NOUN fis-4216 108 11 brings bring VERB fis-4216 108 12 in in ADP fis-4216 108 13 non non ADJ fis-4216 108 14 - - ADJ fis-4216 108 15 linearity linearity ADJ fis-4216 108 16 and and CCONJ fis-4216 108 17 effectively effectively ADV fis-4216 108 18 speeds speed VERB fis-4216 108 19 up up ADP fis-4216 108 20 the the DET fis-4216 108 21 training training NOUN fis-4216 108 22 and and CCONJ fis-4216 108 23 evaluation evaluation NOUN fis-4216 108 24 phases phase NOUN fis-4216 108 25 of of ADP fis-4216 108 26 the the DET fis-4216 108 27 network network NOUN fis-4216 108 28 [ [ X fis-4216 108 29 20 20 NUM fis-4216 108 30 ] ] PUNCT fis-4216 108 31 . . PUNCT fis-4216 109 1 the the DET fis-4216 109 2 concatenate concatenate PROPN fis-4216 109 3 operation operation NOUN fis-4216 109 4 then then ADV fis-4216 109 5 stacks stack VERB fis-4216 109 6 the the DET fis-4216 109 7 expanded expand VERB fis-4216 109 8 outputs output NOUN fis-4216 109 9 up up ADP fis-4216 109 10 in in ADP fis-4216 109 11 the the DET fis-4216 109 12 depth depth NOUN fis-4216 109 13 dimension dimension NOUN fis-4216 109 14 as as ADP fis-4216 109 15 the the DET fis-4216 109 16 input input NOUN fis-4216 109 17 tensor tensor NOUN fis-4216 109 18 of of ADP fis-4216 109 19 the the DET fis-4216 109 20 subsequent subsequent ADJ fis-4216 109 21 convolution convolution NOUN fis-4216 109 22 layer layer NOUN fis-4216 109 23 . . PUNCT fis-4216 110 1 max max PROPN fis-4216 110 2 - - PUNCT fis-4216 110 3 pooling pooling NOUN fis-4216 110 4 with with ADP fis-4216 110 5 a a DET fis-4216 110 6 stride stride ADJ fis-4216 110 7 step step NOUN fis-4216 110 8 size size NOUN fis-4216 110 9 of of ADP fis-4216 110 10 2 2 NUM fis-4216 110 11 was be AUX fis-4216 110 12 performed perform VERB fis-4216 110 13 after after ADP fis-4216 110 14 the the DET fis-4216 110 15 conv1 conv1 NOUN fis-4216 110 16 , , PUNCT fis-4216 110 17 fire3 fire3 PROPN fis-4216 110 18 , , PUNCT fis-4216 110 19 and and CCONJ fis-4216 110 20 fire5 fire5 NOUN fis-4216 110 21 layers layer NOUN fis-4216 110 22 to to PART fis-4216 110 23 summarize summarize VERB fis-4216 110 24 the the DET fis-4216 110 25 feature feature NOUN fis-4216 110 26 response response NOUN fis-4216 110 27 across across ADP fis-4216 110 28 neighboring neighboring NOUN fis-4216 110 29 pixels pixel NOUN fis-4216 110 30 . . PUNCT fis-4216 111 1 the the DET fis-4216 111 2 max max PROPN fis-4216 111 3 - - PUNCT fis-4216 111 4 pooling pool VERB fis-4216 111 5 process process NOUN fis-4216 111 6 with with ADP fis-4216 111 7 a a DET fis-4216 111 8 pooling pool VERB fis-4216 111 9 filter filter NOUN fis-4216 111 10 size size NOUN fis-4216 111 11 of of ADP fis-4216 111 12 3x3 3x3 NUM fis-4216 111 13 in in ADP fis-4216 111 14 the the DET fis-4216 111 15 presented present VERB fis-4216 111 16 model model NOUN fis-4216 111 17 further far ADV fis-4216 111 18 reduces reduce VERB fis-4216 111 19 the the DET fis-4216 111 20 feature feature NOUN fis-4216 111 21 map map NOUN fis-4216 111 22 y y NOUN fis-4216 111 23 , , PUNCT fis-4216 111 24 negative negative ADJ fis-4216 111 25 positive positive ADJ fis-4216 111 26 positive positive ADJ fis-4216 111 27 positive positive ADJ fis-4216 111 28 positive positive ADJ fis-4216 111 29 rotating rotating NOUN fis-4216 111 30 flipping flip VERB fis-4216 111 31 l. l. PROPN fis-4216 111 32 wang wang PROPN fis-4216 111 33 , , PUNCT fis-4216 111 34 frattura frattura PROPN fis-4216 111 35 ed ed PROPN fis-4216 111 36 integrità integrità PROPN fis-4216 111 37 strutturale strutturale PROPN fis-4216 111 38 , , PUNCT fis-4216 111 39 65 65 NUM fis-4216 111 40 ( ( PUNCT fis-4216 111 41 2023 2023 NUM fis-4216 111 42 ) ) PUNCT fis-4216 111 43 289 289 NUM fis-4216 111 44 - - SYM fis-4216 111 45 299 299 NUM fis-4216 111 46 ; ; PUNCT fis-4216 111 47 doi doi NOUN fis-4216 111 48 : : PUNCT fis-4216 111 49 10.3221 10.3221 NUM fis-4216 111 50 / / SYM fis-4216 111 51 igf igf PROPN fis-4216 111 52 - - PUNCT fis-4216 111 53 esis.65.19 esis.65.19 NOUN fis-4216 111 54 293 293 NUM fis-4216 111 55 i i NOUN fis-4216 111 56 , , PUNCT fis-4216 111 57 j j PROPN fis-4216 111 58 i+t i+t NUM fis-4216 111 59 , , PUNCT fis-4216 111 60 j+r j+r PROPN fis-4216 111 61     SPACE fis-4216 111 62 z z PROPN fis-4216 112 1 = = NOUN fis-4216 112 2 max(y max(y X fis-4216 112 3 ) ) PUNCT fis-4216 112 4     SPACE fis-4216 112 5 with with ADP fis-4216 112 6 t t PROPN fis-4216 112 7 , , PUNCT fis-4216 112 8 r r NOUN fis-4216 112 9 = = NOUN fis-4216 112 10 0,1,2 0,1,2 NOUN fis-4216 112 11 . . PUNCT fis-4216 113 1 ( ( PUNCT fis-4216 113 2 2 2 X fis-4216 113 3 ) ) PUNCT fis-4216 113 4 these these DET fis-4216 113 5 processes process NOUN fis-4216 113 6 allow allow VERB fis-4216 113 7 the the DET fis-4216 113 8 implemented implement VERB fis-4216 113 9 algorithm algorithm NOUN fis-4216 113 10 to to PART fis-4216 113 11 learn learn VERB fis-4216 113 12 spatially spatially ADV fis-4216 113 13 invariant invariant ADJ fis-4216 113 14 features feature NOUN fis-4216 113 15 . . PUNCT fis-4216 114 1 figure figure VERB fis-4216 114 2 3 3 NUM fis-4216 114 3 : : PUNCT fis-4216 114 4 schematic schematic ADJ fis-4216 114 5 illustration illustration NOUN fis-4216 114 6 of of ADP fis-4216 114 7 the the DET fis-4216 114 8 2d 2d NUM fis-4216 114 9 3 3 NUM fis-4216 114 10 × × NOUN fis-4216 114 11 3 3 NUM fis-4216 114 12 filter filter NOUN fis-4216 114 13 and and CCONJ fis-4216 114 14 a a DET fis-4216 114 15 stride stride NOUN fis-4216 114 16 of of ADP fis-4216 114 17 2 2 NUM fis-4216 114 18 , , PUNCT fis-4216 114 19 and and CCONJ fis-4216 114 20 a a DET fis-4216 114 21 max max PROPN fis-4216 114 22 pooling pool VERB fis-4216 114 23 with with ADP fis-4216 114 24 size size NOUN fis-4216 114 25 3 3 NUM fis-4216 114 26 × × NOUN fis-4216 114 27 3 3 NUM fis-4216 114 28 and and CCONJ fis-4216 114 29 a a DET fis-4216 114 30 stride stride NOUN fis-4216 114 31 of of ADP fis-4216 114 32 2 2 NUM fis-4216 114 33 . . PUNCT fis-4216 115 1 the the DET fis-4216 115 2 network network NOUN fis-4216 115 3 training training NOUN fis-4216 115 4 aims aim VERB fis-4216 115 5 to to PART fis-4216 115 6 increase increase VERB fis-4216 115 7 the the DET fis-4216 115 8 variation variation NOUN fis-4216 115 9 extracted extract VERB fis-4216 115 10 from from ADP fis-4216 115 11 the the DET fis-4216 115 12 input input NOUN fis-4216 115 13 image image NOUN fis-4216 115 14 data datum NOUN fis-4216 115 15 while while SCONJ fis-4216 115 16 preventing prevent VERB fis-4216 115 17 overfitting overfitting NOUN fis-4216 115 18 . . PUNCT fis-4216 116 1 a a DET fis-4216 116 2 dropout dropout NOUN fis-4216 116 3 action action NOUN fis-4216 116 4 with with ADP fis-4216 116 5 a a DET fis-4216 116 6 probability probability NOUN fis-4216 116 7 of of ADP fis-4216 116 8 0.5 0.5 NUM fis-4216 116 9 prevents prevent NOUN fis-4216 116 10 complex complex ADJ fis-4216 116 11 co co NOUN fis-4216 116 12 - - NOUN fis-4216 116 13 adaptations adaptation NOUN fis-4216 116 14 after after ADP fis-4216 116 15 the the DET fis-4216 116 16 fire9 fire9 PROPN fis-4216 116 17 module module NOUN fis-4216 116 18 , , PUNCT fis-4216 116 19 reducing reduce VERB fis-4216 116 20 the the DET fis-4216 116 21 possibility possibility NOUN fis-4216 116 22 of of ADP fis-4216 116 23 overfitting overfitte VERB fis-4216 116 24 . . PUNCT fis-4216 117 1 the the DET fis-4216 117 2 global global ADJ fis-4216 117 3 average average ADJ fis-4216 117 4 pooling pooling NOUN fis-4216 117 5 operation operation NOUN fis-4216 117 6 converts convert VERB fis-4216 117 7 the the DET fis-4216 117 8 class class NOUN fis-4216 117 9 feature feature NOUN fis-4216 117 10 maps map NOUN fis-4216 117 11 into into ADP fis-4216 117 12 a a DET fis-4216 117 13 single single ADJ fis-4216 117 14 value value NOUN fis-4216 117 15 later later ADV fis-4216 117 16 . . PUNCT fis-4216 118 1 a a DET fis-4216 118 2 full full ADV fis-4216 118 3 - - PUNCT fis-4216 118 4 connected connect VERB fis-4216 118 5 2 2 NUM fis-4216 118 6 - - PUNCT fis-4216 118 7 class class NOUN fis-4216 118 8 layer layer NOUN fis-4216 118 9 was be AUX fis-4216 118 10 implemented implement VERB fis-4216 118 11 to to PART fis-4216 118 12 replace replace VERB fis-4216 118 13 the the DET fis-4216 118 14 original original ADJ fis-4216 118 15 1000 1000 NUM fis-4216 118 16 - - PUNCT fis-4216 118 17 class class NOUN fis-4216 118 18 layer layer NOUN fis-4216 118 19 of of ADP fis-4216 118 20 squeezenet squeezenet NOUN fis-4216 118 21 [ [ X fis-4216 118 22 21 21 NUM fis-4216 118 23 ] ] PUNCT fis-4216 118 24 . . PUNCT fis-4216 119 1 at at ADP fis-4216 119 2 the the DET fis-4216 119 3 end end NOUN fis-4216 119 4 of of ADP fis-4216 119 5 the the DET fis-4216 119 6 network network NOUN fis-4216 119 7 , , PUNCT fis-4216 119 8 the the DET fis-4216 119 9 softmax softmax NOUN fis-4216 119 10 activation activation NOUN fis-4216 119 11 function function NOUN fis-4216 119 12 calculates calculate VERB fis-4216 119 13 the the DET fis-4216 119 14 probability probability NOUN fis-4216 119 15 of of ADP fis-4216 119 16 negative negative ADJ fis-4216 119 17 or or CCONJ fis-4216 119 18 positive positive ADJ fis-4216 119 19 classes class NOUN fis-4216 119 20 given give VERB fis-4216 119 21 an an DET fis-4216 119 22 input input NOUN fis-4216 119 23 image image NOUN fis-4216 119 24 for for ADP fis-4216 119 25 the the DET fis-4216 119 26 final final ADJ fis-4216 119 27 classification classification NOUN fis-4216 119 28 decision decision NOUN fis-4216 119 29 - - PUNCT fis-4216 119 30 making making NOUN fis-4216 119 31 . . PUNCT fis-4216 120 1 given give VERB fis-4216 120 2 an an DET fis-4216 120 3 image image NOUN fis-4216 120 4 dataset dataset NOUN fis-4216 120 5 s={p(i s={p(i NOUN fis-4216 120 6 ) ) PUNCT fis-4216 120 7 , , PUNCT fis-4216 120 8 l(i l(i NOUN fis-4216 120 9 ) ) PUNCT fis-4216 120 10 } } PUNCT fis-4216 120 11 of of ADP fis-4216 120 12 m m PROPN fis-4216 120 13 image image NOUN fis-4216 120 14 patches patch NOUN fis-4216 120 15 , , PUNCT fis-4216 120 16 where where SCONJ fis-4216 120 17 p(i p(i PROPN fis-4216 120 18 ) ) PUNCT fis-4216 120 19 is be AUX fis-4216 120 20 the the DET fis-4216 120 21 i i PROPN fis-4216 120 22 - - PUNCT fis-4216 120 23 th th X fis-4216 120 24 patch patch NOUN fis-4216 120 25 and and CCONJ fis-4216 120 26 l(i l(i NOUN fis-4216 120 27 ) ) PUNCT fis-4216 120 28 ∈{0,1 ∈{0,1 NOUN fis-4216 120 29 } } PUNCT fis-4216 120 30 is be AUX fis-4216 120 31 the the DET fis-4216 120 32 associated associated ADJ fis-4216 120 33 class class NOUN fis-4216 120 34 label label NOUN fis-4216 120 35 . . PUNCT fis-4216 121 1 if if SCONJ fis-4216 121 2 p(i p(i PROPN fis-4216 121 3 ) ) PUNCT fis-4216 121 4 is be AUX fis-4216 121 5 a a DET fis-4216 121 6 positive positive ADJ fis-4216 121 7 patch patch NOUN fis-4216 121 8 , , PUNCT fis-4216 121 9 l(i l(i NOUN fis-4216 121 10 ) ) PUNCT fis-4216 121 11 is be AUX fis-4216 121 12 assigned assign VERB fis-4216 121 13 as as ADP fis-4216 121 14 1 1 NUM fis-4216 121 15 ; ; PUNCT fis-4216 121 16 otherwise otherwise ADV fis-4216 121 17 , , PUNCT fis-4216 121 18 l(i l(i NOUN fis-4216 121 19 ) ) PUNCT fis-4216 121 20 is be AUX fis-4216 121 21 set set VERB fis-4216 121 22 to to ADP fis-4216 121 23 0 0 NUM fis-4216 121 24 . . PUNCT fis-4216 122 1 let let VERB fis-4216 122 2 ( ( PUNCT fis-4216 122 3 i i NOUN fis-4216 122 4 ) ) PUNCT fis-4216 122 5 jz jz PROPN fis-4216 122 6 be be VERB fis-4216 122 7 the the DET fis-4216 122 8 output output NOUN fis-4216 122 9 of of ADP fis-4216 122 10 unit unit NOUN fis-4216 122 11 j j PROPN fis-4216 122 12 in in ADP fis-4216 122 13 the the DET fis-4216 122 14 softmax softmax NOUN fis-4216 122 15 layer layer NOUN fis-4216 122 16 for for ADP fis-4216 122 17 image image NOUN fis-4216 122 18 p(i p(i PROPN fis-4216 122 19 ) ) PUNCT fis-4216 122 20 , , PUNCT fis-4216 122 21 then then ADV fis-4216 122 22 the the DET fis-4216 122 23 probability probability NOUN fis-4216 122 24 that that SCONJ fis-4216 122 25 the the DET fis-4216 122 26 label label NOUN fis-4216 122 27 l(i l(i NOUN fis-4216 122 28 ) ) PUNCT fis-4216 122 29 of of ADP fis-4216 122 30 p(i p(i PROPN fis-4216 122 31 ) ) PUNCT fis-4216 122 32 is be AUX fis-4216 122 33 j j PROPN fis-4216 122 34 can can AUX fis-4216 122 35 be be AUX fis-4216 122 36 calculated calculate VERB fis-4216 122 37 by by ADP fis-4216 122 38 the the DET fis-4216 122 39 formula formula NOUN fis-4216 122 40 ∑ ∑ PUNCT fis-4216 122 41 ( ( PUNCT fis-4216 122 42 i i NOUN fis-4216 122 43 ) ) PUNCT fis-4216 122 44 j j PROPN fis-4216 122 45 ( ( PUNCT fis-4216 122 46 i i NOUN fis-4216 122 47 ) ) PUNCT fis-4216 122 48 l l PROPN fis-4216 123 1 z z NOUN fis-4216 123 2 ( ( PUNCT fis-4216 123 3 i i NOUN fis-4216 123 4 ) ) PUNCT fis-4216 123 5 ( ( PUNCT fis-4216 123 6 i i NOUN fis-4216 123 7 ) ) PUNCT fis-4216 123 8 j j PROPN fis-4216 124 1 k k NOUN fis-4216 124 2 z z PUNCT fis-4216 125 1 l=1 l=1 X fis-4216 125 2 e e X fis-4216 125 3 p(l p(l NOUN fis-4216 125 4 = = NOUN fis-4216 125 5 j|z j|z NOUN fis-4216 125 6 ) ) PUNCT fis-4216 126 1 = = SYM fis-4216 126 2 e e X fis-4216 126 3 ( ( PUNCT fis-4216 126 4 3 3 X fis-4216 126 5 ) ) PUNCT fis-4216 126 6 the the DET fis-4216 126 7 literature literature NOUN fis-4216 126 8 [ [ X fis-4216 126 9 19 19 NUM fis-4216 126 10 ] ] PUNCT fis-4216 126 11 is be AUX fis-4216 126 12 referred refer VERB fis-4216 126 13 to to ADP fis-4216 126 14 interested interested ADJ fis-4216 126 15 readers reader NOUN fis-4216 126 16 for for ADP fis-4216 126 17 algorithm algorithm NOUN fis-4216 126 18 details detail NOUN fis-4216 126 19 . . PUNCT fis-4216 127 1 during during ADP fis-4216 127 2 training training NOUN fis-4216 127 3 , , PUNCT fis-4216 127 4 the the DET fis-4216 127 5 first first ADJ fis-4216 127 6 small small ADJ fis-4216 127 7 batch batch NOUN fis-4216 127 8 images image NOUN fis-4216 127 9 of of ADP fis-4216 127 10 concrete concrete ADJ fis-4216 127 11 surface surface NOUN fis-4216 127 12 were be AUX fis-4216 127 13 input input VERB fis-4216 127 14 into into ADP fis-4216 127 15 the the DET fis-4216 127 16 squeezenet squeezenet NOUN fis-4216 127 17 with with ADP fis-4216 127 18 initial initial ADJ fis-4216 127 19 parameter parameter NOUN fis-4216 127 20 settings setting NOUN fis-4216 127 21 . . PUNCT fis-4216 128 1 in in ADP fis-4216 128 2 the the DET fis-4216 128 3 squeezenet squeezenet NOUN fis-4216 128 4 network network NOUN fis-4216 128 5 , , PUNCT fis-4216 128 6 max max PROPN fis-4216 128 7 pool pool NOUN fis-4216 128 8 layers layer NOUN fis-4216 128 9 execute execute VERB fis-4216 128 10 a a DET fis-4216 128 11 down down ADV fis-4216 128 12 - - PUNCT fis-4216 128 13 sampling sample VERB fis-4216 128 14 operation operation NOUN fis-4216 128 15 in in ADP fis-4216 128 16 spatial spatial ADJ fis-4216 128 17 dimensions dimension NOUN fis-4216 128 18 . . PUNCT fis-4216 129 1 the the DET fis-4216 129 2 fire fire NOUN fis-4216 129 3 moduli modulus NOUN fis-4216 129 4 sequentially sequentially ADV fis-4216 129 5 transform transform VERB fis-4216 129 6 the the DET fis-4216 129 7 image image NOUN fis-4216 129 8 inputs input NOUN fis-4216 129 9 into into ADP fis-4216 129 10 high high ADJ fis-4216 129 11 - - PUNCT fis-4216 129 12 level level NOUN fis-4216 129 13 feature feature NOUN fis-4216 129 14 maps map NOUN fis-4216 129 15 through through ADP fis-4216 129 16 the the DET fis-4216 129 17 squeeze squeeze NOUN fis-4216 129 18 and and CCONJ fis-4216 129 19 expand expand VERB fis-4216 129 20 operations operation NOUN fis-4216 129 21 . . PUNCT fis-4216 130 1 the the DET fis-4216 130 2 network network NOUN fis-4216 130 3 outputs output NOUN fis-4216 130 4 are be AUX fis-4216 130 5 a a DET fis-4216 130 6 weighted weight VERB fis-4216 130 7 combination combination NOUN fis-4216 130 8 of of ADP fis-4216 130 9 the the DET fis-4216 130 10 feature feature NOUN fis-4216 130 11 maps map NOUN fis-4216 130 12 on on ADP fis-4216 130 13 tensors tensor NOUN fis-4216 130 14 . . PUNCT fis-4216 131 1 the the DET fis-4216 131 2 feature feature NOUN fis-4216 131 3 map map NOUN fis-4216 131 4 of of ADP fis-4216 131 5 classes class NOUN fis-4216 131 6 is be AUX fis-4216 131 7 later later ADV fis-4216 131 8 converted convert VERB fis-4216 131 9 into into ADP fis-4216 131 10 one one NUM fis-4216 131 11 value value NOUN fis-4216 131 12 by by ADP fis-4216 131 13 the the DET fis-4216 131 14 global global ADJ fis-4216 131 15 average average ADJ fis-4216 131 16 pool pool NOUN fis-4216 131 17 operation operation NOUN fis-4216 131 18 , , PUNCT fis-4216 131 19 which which PRON fis-4216 131 20 gives give VERB fis-4216 131 21 the the DET fis-4216 131 22 multiclass multiclass ADJ fis-4216 131 23 probability probability NOUN fis-4216 131 24 distribution distribution NOUN fis-4216 131 25 by by ADP fis-4216 131 26 the the DET fis-4216 131 27 softmax softmax NOUN fis-4216 131 28 activation activation NOUN fis-4216 131 29 function function NOUN fis-4216 131 30 at at ADP fis-4216 131 31 the the DET fis-4216 131 32 end end NOUN fis-4216 131 33 of of ADP fis-4216 131 34 the the DET fis-4216 131 35 network network NOUN fis-4216 131 36 . . PUNCT fis-4216 132 1 the the DET fis-4216 132 2 loss loss NOUN fis-4216 132 3 function function NOUN fis-4216 132 4 in in ADP fis-4216 132 5 the the DET fis-4216 132 6 cross cross NOUN fis-4216 132 7 entropy entropy PROPN fis-4216 132 8 loss loss NOUN fis-4216 132 9 is be AUX fis-4216 132 10 calculated calculate VERB fis-4216 132 11 to to PART fis-4216 132 12 compare compare VERB fis-4216 132 13 the the DET fis-4216 132 14 classification classification NOUN fis-4216 132 15 results result NOUN fis-4216 132 16 to to ADP fis-4216 132 17 image image NOUN fis-4216 132 18 labels label NOUN fis-4216 132 19 . . PUNCT fis-4216 133 1 the the DET fis-4216 133 2 parameters parameter NOUN fis-4216 133 3 , , PUNCT fis-4216 133 4 weighting weighting NOUN fis-4216 133 5 , , PUNCT fis-4216 133 6 and and CCONJ fis-4216 133 7 functionality functionality NOUN fis-4216 133 8 are be AUX fis-4216 133 9 then then ADV fis-4216 133 10 adjusted adjust VERB fis-4216 133 11 , , PUNCT fis-4216 133 12 as as SCONJ fis-4216 133 13 needed need VERB fis-4216 133 14 , , PUNCT fis-4216 133 15 for for ADP fis-4216 133 16 network network NOUN fis-4216 133 17 training training NOUN fis-4216 133 18 on on ADP fis-4216 133 19 the the DET fis-4216 133 20 following follow VERB fis-4216 133 21 batch batch NOUN fis-4216 133 22 images image NOUN fis-4216 133 23 iteratively iteratively ADV fis-4216 133 24 . . PUNCT fis-4216 134 1 in in ADP fis-4216 134 2 parallel parallel NOUN fis-4216 134 3 , , PUNCT fis-4216 134 4 a a DET fis-4216 134 5 validation validation NOUN fis-4216 134 6 process process NOUN fis-4216 134 7 is be AUX fis-4216 134 8 performed perform VERB fis-4216 134 9 to to PART fis-4216 134 10 estimate estimate VERB fis-4216 134 11 the the DET fis-4216 134 12 accuracy accuracy NOUN fis-4216 134 13 of of ADP fis-4216 134 14 the the DET fis-4216 134 15 developing develop VERB fis-4216 134 16 network network NOUN fis-4216 134 17 . . PUNCT fis-4216 135 1 the the DET fis-4216 135 2 validation validation NOUN fis-4216 135 3 process process NOUN fis-4216 135 4 aims aim VERB fis-4216 135 5 to to PART fis-4216 135 6 assess assess VERB fis-4216 135 7 the the DET fis-4216 135 8 function function NOUN fis-4216 135 9 and and CCONJ fis-4216 135 10 performance performance NOUN fis-4216 135 11 in in ADP fis-4216 135 12 crack crack NOUN fis-4216 135 13 detection detection NOUN fis-4216 135 14 of of ADP fis-4216 135 15 the the DET fis-4216 135 16 squeezenet squeezenet NOUN fis-4216 135 17 under under ADP fis-4216 135 18 training training NOUN fis-4216 135 19 against against ADP fis-4216 135 20 unbiased unbiased ADJ fis-4216 135 21 and and CCONJ fis-4216 135 22 independent independent ADJ fis-4216 135 23 inputs input NOUN fis-4216 135 24 . . PUNCT fis-4216 136 1 the the DET fis-4216 136 2 hyperparameters hyperparameter NOUN fis-4216 136 3 controlling control VERB fis-4216 136 4 the the DET fis-4216 136 5 overall overall ADJ fis-4216 136 6 process process NOUN fis-4216 136 7 are be AUX fis-4216 136 8 updated update VERB fis-4216 136 9 if if SCONJ fis-4216 136 10 the the DET fis-4216 136 11 validation validation NOUN fis-4216 136 12 error error NOUN fis-4216 136 13 is be AUX fis-4216 136 14 unsatisfactory unsatisfactory ADJ fis-4216 136 15 . . PUNCT fis-4216 137 1 after after ADP fis-4216 137 2 various various ADJ fis-4216 137 3 iterations iteration NOUN fis-4216 137 4 of of ADP fis-4216 137 5 resampling resample VERB fis-4216 137 6 and and CCONJ fis-4216 137 7 fine fine ADV fis-4216 137 8 - - PUNCT fis-4216 137 9 tuning tuning NOUN fis-4216 137 10 in in ADP fis-4216 137 11 train train NOUN fis-4216 137 12 and and CCONJ fis-4216 137 13 validation validation NOUN fis-4216 137 14 stages stage NOUN fis-4216 137 15 , , PUNCT fis-4216 137 16 the the DET fis-4216 137 17 best good ADJ fis-4216 137 18 squeezenet squeezenet NOUN fis-4216 137 19 model model NOUN fis-4216 137 20 is be AUX fis-4216 137 21 obtained obtain VERB fis-4216 137 22 once once ADV fis-4216 137 23 the the DET fis-4216 137 24 desired desire VERB fis-4216 137 25 performance performance NOUN fis-4216 137 26 metrics metric NOUN fis-4216 137 27 are be AUX fis-4216 137 28 achieved achieve VERB fis-4216 137 29 . . PUNCT fis-4216 138 1 the the DET fis-4216 138 2 functionality functionality NOUN fis-4216 138 3 of of ADP fis-4216 138 4 the the DET fis-4216 138 5 best good ADJ fis-4216 138 6 model model NOUN fis-4216 138 7 obtained obtain VERB fis-4216 138 8 is be AUX fis-4216 138 9 verified verify VERB fis-4216 138 10 on on ADP fis-4216 138 11 the the DET fis-4216 138 12 test test NOUN fis-4216 138 13 or or CCONJ fis-4216 138 14 l. l. PROPN fis-4216 138 15 wang wang PROPN fis-4216 138 16 , , PUNCT fis-4216 138 17 frattura frattura PROPN fis-4216 138 18 ed ed PROPN fis-4216 138 19 integrità integrità PROPN fis-4216 138 20 strutturale strutturale PROPN fis-4216 138 21 , , PUNCT fis-4216 138 22 65 65 NUM fis-4216 138 23 ( ( PUNCT fis-4216 138 24 2023 2023 NUM fis-4216 138 25 ) ) PUNCT fis-4216 138 26 289 289 NUM fis-4216 138 27 - - SYM fis-4216 138 28 299 299 NUM fis-4216 138 29 ; ; PUNCT fis-4216 138 30 doi doi NOUN fis-4216 138 31 : : PUNCT fis-4216 138 32 10.3221 10.3221 NUM fis-4216 138 33 / / SYM fis-4216 138 34 igf igf PROPN fis-4216 138 35 - - PUNCT fis-4216 138 36 esis.65.19 esis.65.19 NOUN fis-4216 138 37 294 294 NUM fis-4216 138 38 holdout holdout NOUN fis-4216 138 39 image image NOUN fis-4216 138 40 set set VERB fis-4216 138 41 for for ADP fis-4216 138 42 practical practical ADJ fis-4216 138 43 application application NOUN fis-4216 138 44 on on ADP fis-4216 138 45 unseen unseen ADJ fis-4216 138 46 inputs input NOUN fis-4216 138 47 . . PUNCT fis-4216 139 1 in in ADP fis-4216 139 2 the the DET fis-4216 139 3 test test NOUN fis-4216 139 4 procedure procedure NOUN fis-4216 139 5 , , PUNCT fis-4216 139 6 the the DET fis-4216 139 7 model model NOUN fis-4216 139 8 works work VERB fis-4216 139 9 as as ADP fis-4216 139 10 a a DET fis-4216 139 11 black black ADJ fis-4216 139 12 box box NOUN fis-4216 139 13 with with ADP fis-4216 139 14 images image NOUN fis-4216 139 15 passed pass VERB fis-4216 139 16 through through ADP fis-4216 139 17 it it PRON fis-4216 139 18 to to PART fis-4216 139 19 diagnose diagnose VERB fis-4216 139 20 if if SCONJ fis-4216 139 21 a a DET fis-4216 139 22 crack crack NOUN fis-4216 139 23 exists exist VERB fis-4216 139 24 in in ADP fis-4216 139 25 the the DET fis-4216 139 26 concrete concrete ADJ fis-4216 139 27 surface surface NOUN fis-4216 139 28 or or CCONJ fis-4216 139 29 not not PART fis-4216 139 30 . . PUNCT fis-4216 140 1 the the DET fis-4216 140 2 hyperparameters hyperparameter NOUN fis-4216 140 3 of of ADP fis-4216 140 4 the the DET fis-4216 140 5 model model NOUN fis-4216 140 6 will will AUX fis-4216 140 7 not not PART fis-4216 140 8 be be AUX fis-4216 140 9 adjusted adjust VERB fis-4216 140 10 in in ADP fis-4216 140 11 the the DET fis-4216 140 12 testing testing NOUN fis-4216 140 13 stage stage NOUN fis-4216 140 14 . . PUNCT fis-4216 141 1 layer layer NOUN fis-4216 141 2 type type NOUN fis-4216 141 3 kernel kernel PROPN fis-4216 141 4 size size NOUN fis-4216 141 5 stride stride NOUN fis-4216 141 6 padding padding NOUN fis-4216 141 7 output output NOUN fis-4216 141 8 size size NOUN fis-4216 141 9 data data VERB fis-4216 141 10 227×227×3 227×227×3 NUM fis-4216 141 11 conv1 conv1 NOUN fis-4216 141 12 { { PUNCT fis-4216 141 13 conv+relu conv+relu PROPN fis-4216 141 14 } } PUNCT fis-4216 141 15 3×3 3×3 NUM fis-4216 141 16 2 2 NUM fis-4216 141 17 0 0 NUM fis-4216 141 18 113×113×64 113×113×64 NUM fis-4216 141 19 pool1 pool1 NOUN fis-4216 141 20 max max PROPN fis-4216 141 21 pool pool NOUN fis-4216 141 22 3×3 3×3 NUM fis-4216 141 23 2 2 NUM fis-4216 141 24 0 0 NUM fis-4216 141 25 56×56×64 56×56×64 NUM fis-4216 141 26 fire2 fire2 PROPN fis-4216 141 27 / / SYM fis-4216 141 28 fire3 fire3 PROPN fis-4216 141 29 { { PUNCT fis-4216 141 30 squeeze+relu squeeze+relu PROPN fis-4216 141 31 } } PUNCT fis-4216 141 32 1×1 1×1 NOUN fis-4216 141 33 1 1 NUM fis-4216 141 34 0 0 NUM fis-4216 141 35 56×56×16 56×56×16 NUM fis-4216 141 36 { { PUNCT fis-4216 141 37 expand+relu expand+relu NOUN fis-4216 141 38 } } PUNCT fis-4216 141 39 1×1 1×1 NOUN fis-4216 141 40 1 1 NUM fis-4216 141 41 0 0 NUM fis-4216 142 1 56×56×64 56×56×64 NUM fis-4216 142 2 { { PUNCT fis-4216 142 3 expand+relu expand+relu PROPN fis-4216 142 4 } } PUNCT fis-4216 142 5 3×3 3×3 NUM fis-4216 142 6 1 1 NUM fis-4216 142 7 1 1 NUM fis-4216 142 8 56×56×64 56×56×64 NUM fis-4216 142 9 concat concat NOUN fis-4216 142 10 56×56×128 56×56×128 NUM fis-4216 142 11 pool3 pool3 NOUN fis-4216 142 12 max max NOUN fis-4216 142 13 pool pool NOUN fis-4216 142 14 3×3 3×3 NUM fis-4216 142 15 2 2 NUM fis-4216 142 16 0 0 NUM fis-4216 142 17 28×28×128 28×28×128 NUM fis-4216 142 18 fire4 fire4 ADJ fis-4216 142 19 / / SYM fis-4216 142 20 fire5 fire5 NOUN fis-4216 142 21 { { PUNCT fis-4216 142 22 squeeze+relu squeeze+relu PROPN fis-4216 142 23 } } PUNCT fis-4216 142 24 1×1 1×1 NOUN fis-4216 142 25 1 1 NUM fis-4216 142 26 0 0 NUM fis-4216 142 27 28×28×32 28×28×32 NUM fis-4216 142 28 { { PUNCT fis-4216 142 29 expand+relu expand+relu NOUN fis-4216 142 30 } } PUNCT fis-4216 142 31 1×1 1×1 NOUN fis-4216 142 32 1 1 NUM fis-4216 142 33 0 0 NUM fis-4216 142 34 28×28×128 28×28×128 NUM fis-4216 142 35 { { PUNCT fis-4216 142 36 expand+relu expand+relu PROPN fis-4216 142 37 } } PUNCT fis-4216 142 38 3×3 3×3 NUM fis-4216 142 39 1 1 NUM fis-4216 142 40 1 1 NUM fis-4216 142 41 28×28×128 28×28×128 NUM fis-4216 142 42 concat concat NOUN fis-4216 142 43 28×28×256 28×28×256 NUM fis-4216 142 44 pool5 pool5 NOUN fis-4216 142 45 max max PROPN fis-4216 142 46 pool pool NOUN fis-4216 142 47 3×3 3×3 NUM fis-4216 142 48 2 2 NUM fis-4216 142 49 0 0 NUM fis-4216 142 50 14×14×256 14×14×256 NUM fis-4216 142 51 fire6 fire6 NOUN fis-4216 142 52 / / SYM fis-4216 142 53 fire7 fire7 NOUN fis-4216 142 54 { { PUNCT fis-4216 142 55 squeeze+relu squeeze+relu PROPN fis-4216 142 56 } } PUNCT fis-4216 142 57 1×1 1×1 NOUN fis-4216 142 58 1 1 NUM fis-4216 142 59 0 0 NUM fis-4216 142 60 14×14×48 14×14×48 NUM fis-4216 142 61 { { PUNCT fis-4216 142 62 expand+relu expand+relu NOUN fis-4216 142 63 } } PUNCT fis-4216 142 64 1×1 1×1 NOUN fis-4216 142 65 1 1 NUM fis-4216 142 66 0 0 NUM fis-4216 142 67 14×14×192 14×14×192 NUM fis-4216 142 68 { { PUNCT fis-4216 142 69 expand+relu expand+relu PROPN fis-4216 142 70 } } PUNCT fis-4216 142 71 3×3 3×3 NUM fis-4216 142 72 1 1 NUM fis-4216 142 73 1 1 NUM fis-4216 142 74 14×14×192 14×14×192 NUM fis-4216 142 75 concat concat NOUN fis-4216 142 76 14×14×384 14×14×384 NUM fis-4216 142 77 fire8 fire8 NOUN fis-4216 142 78 / / SYM fis-4216 142 79 fire9 fire9 PROPN fis-4216 142 80 { { PUNCT fis-4216 142 81 squeeze+relu squeeze+relu PROPN fis-4216 142 82 } } PUNCT fis-4216 142 83 1×1 1×1 NOUN fis-4216 142 84 1 1 NUM fis-4216 142 85 0 0 NUM fis-4216 142 86 14×14×64 14×14×64 NUM fis-4216 142 87 { { PUNCT fis-4216 142 88 expand+relu expand+relu NOUN fis-4216 142 89 } } PUNCT fis-4216 142 90 1×1 1×1 NOUN fis-4216 142 91 1 1 NUM fis-4216 142 92 0 0 NUM fis-4216 142 93 14×14×256 14×14×256 NUM fis-4216 142 94 { { PUNCT fis-4216 142 95 expand+relu expand+relu PROPN fis-4216 142 96 } } PUNCT fis-4216 142 97 3×3 3×3 NUM fis-4216 142 98 1 1 NUM fis-4216 142 99 1 1 NUM fis-4216 142 100 14×14×256 14×14×256 NUM fis-4216 142 101 concat concat NOUN fis-4216 142 102 14×14×512 14×14×512 PROPN fis-4216 142 103 conv10 conv10 NOUN fis-4216 142 104 conv+relu conv+relu PROPN fis-4216 142 105 1 1 NUM fis-4216 142 106 1 1 NUM fis-4216 142 107 0 0 NUM fis-4216 142 108 14×14×1000 14×14×1000 NUM fis-4216 142 109 pool10 pool10 NOUN fis-4216 142 110 global global ADJ fis-4216 142 111 average average ADJ fis-4216 142 112 pool pool NOUN fis-4216 142 113 1×1×1000 1×1×1000 NUM fis-4216 142 114 fc_add fc_add VERB fis-4216 142 115 fully fully ADV fis-4216 142 116 connect connect VERB fis-4216 142 117 1×1×2 1×1×2 NUM fis-4216 142 118 output output NOUN fis-4216 142 119 softmax softmax NOUN fis-4216 142 120 1×1×2 1×1×2 NUM fis-4216 142 121 table table NOUN fis-4216 142 122 1 1 NUM fis-4216 142 123 : : PUNCT fis-4216 142 124 detailed detailed ADJ fis-4216 142 125 layer layer NOUN fis-4216 142 126 configuration configuration NOUN fis-4216 142 127 of of ADP fis-4216 142 128 the the DET fis-4216 142 129 implemented implement VERB fis-4216 142 130 network network NOUN fis-4216 142 131 . . PUNCT fis-4216 143 1 adam adam PROPN fis-4216 143 2 optimization optimization NOUN fis-4216 143 3 during during ADP fis-4216 143 4 training training NOUN fis-4216 143 5 , , PUNCT fis-4216 143 6 the the DET fis-4216 143 7 parameters parameter NOUN fis-4216 143 8 of of ADP fis-4216 143 9 the the DET fis-4216 143 10 squeezenet squeezenet NOUN fis-4216 143 11 model model NOUN fis-4216 143 12 were be AUX fis-4216 143 13 fine fine ADV fis-4216 143 14 - - PUNCT fis-4216 143 15 tuned tune VERB fis-4216 143 16 through through ADP fis-4216 143 17 the the DET fis-4216 143 18 optimization optimization NOUN fis-4216 143 19 iteration iteration NOUN fis-4216 143 20 updated update VERB fis-4216 143 21 by by ADP fis-4216 143 22 the the DET fis-4216 143 23 validation validation NOUN fis-4216 143 24 error error NOUN fis-4216 143 25 . . PUNCT fis-4216 144 1 the the DET fis-4216 144 2 hyperparameters hyperparameter NOUN fis-4216 144 3 in in ADP fis-4216 144 4 the the DET fis-4216 144 5 deep deep ADJ fis-4216 144 6 learning learning NOUN fis-4216 144 7 model model NOUN fis-4216 144 8 play play VERB fis-4216 144 9 a a DET fis-4216 144 10 crucial crucial ADJ fis-4216 144 11 role role NOUN fis-4216 144 12 in in ADP fis-4216 144 13 deep deep ADJ fis-4216 144 14 learning learning NOUN fis-4216 144 15 algorithms algorithm NOUN fis-4216 144 16 , , PUNCT fis-4216 144 17 as as SCONJ fis-4216 144 18 these these DET fis-4216 144 19 parameters parameter NOUN fis-4216 144 20 tightly tightly ADV fis-4216 144 21 control control VERB fis-4216 144 22 the the DET fis-4216 144 23 actions action NOUN fis-4216 144 24 of of ADP fis-4216 144 25 the the DET fis-4216 144 26 training training NOUN fis-4216 144 27 algorithms algorithm NOUN fis-4216 144 28 and and CCONJ fis-4216 144 29 significantly significantly ADV fis-4216 144 30 affect affect VERB fis-4216 144 31 the the DET fis-4216 144 32 performance performance NOUN fis-4216 144 33 of of ADP fis-4216 144 34 the the DET fis-4216 144 35 models model NOUN fis-4216 144 36 . . PUNCT fis-4216 145 1 the the DET fis-4216 145 2 parameters parameter NOUN fis-4216 145 3 in in ADP fis-4216 145 4 the the DET fis-4216 145 5 squeezenet squeezenet NOUN fis-4216 145 6 model model NOUN fis-4216 145 7 are be AUX fis-4216 145 8 synergistically synergistically ADV fis-4216 145 9 optimized optimize VERB fis-4216 145 10 by by ADP fis-4216 145 11 minimizing minimize VERB fis-4216 145 12 the the DET fis-4216 145 13 misclassification misclassification NOUN fis-4216 145 14 error error NOUN fis-4216 145 15 during during ADP fis-4216 145 16 training training NOUN fis-4216 145 17 . . PUNCT fis-4216 146 1 the the DET fis-4216 146 2 adam adam PROPN fis-4216 146 3 optimization optimization NOUN fis-4216 146 4 algorithm algorithm NOUN fis-4216 146 5 is be AUX fis-4216 146 6 used use VERB fis-4216 146 7 to to ADP fis-4216 146 8 fine fine ADJ fis-4216 146 9 - - PUNCT fis-4216 146 10 tune tune NOUN fis-4216 146 11 the the DET fis-4216 146 12 parameters parameter NOUN fis-4216 146 13 of of ADP fis-4216 146 14 the the DET fis-4216 146 15 squeezenet squeezenet NOUN fis-4216 146 16 - - PUNCT fis-4216 146 17 based base VERB fis-4216 146 18 model model NOUN fis-4216 146 19 . . PUNCT fis-4216 147 1 the the DET fis-4216 147 2 adam adam PROPN fis-4216 147 3 algorithm algorithm PROPN fis-4216 147 4 improves improve VERB fis-4216 147 5 network network NOUN fis-4216 147 6 training training NOUN fis-4216 147 7 by by ADP fis-4216 147 8 using use VERB fis-4216 147 9 learning learn VERB fis-4216 147 10 rates rate NOUN fis-4216 147 11 that that PRON fis-4216 147 12 differ differ VERB fis-4216 147 13 by by ADP fis-4216 147 14 parameter parameter NOUN fis-4216 147 15 and and CCONJ fis-4216 147 16 can can AUX fis-4216 147 17 automatically automatically ADV fis-4216 147 18 adapt adapt VERB fis-4216 147 19 to to ADP fis-4216 147 20 the the DET fis-4216 147 21 loss loss NOUN fis-4216 147 22 function function NOUN fis-4216 147 23 being be AUX fis-4216 147 24 optimized optimize VERB fis-4216 147 25 . . PUNCT fis-4216 148 1 it it PRON fis-4216 148 2 maintains maintain VERB fis-4216 148 3 an an DET fis-4216 148 4 element element NOUN fis-4216 148 5 - - ADJ fis-4216 148 6 wise wise ADJ fis-4216 148 7 moving move VERB fis-4216 148 8 average average NOUN fis-4216 148 9 of of ADP fis-4216 148 10 both both CCONJ fis-4216 148 11 the the DET fis-4216 148 12 parameter parameter NOUN fis-4216 148 13 gradients gradient NOUN fis-4216 148 14 and and CCONJ fis-4216 148 15 their their PRON fis-4216 148 16 squared squared ADJ fis-4216 148 17 values value NOUN fis-4216 148 18 , , PUNCT fis-4216 148 19 ∇l ∇l VERB fis-4216 148 20 1 1 NUM fis-4216 148 21 l-1 l-1 NUM fis-4216 148 22 1 1 NUM fis-4216 148 23 lm lm PROPN fis-4216 148 24 = = NOUN fis-4216 148 25 β β NOUN fis-4216 148 26 m m VERB fis-4216 148 27 + + ADJ fis-4216 148 28 ( ( PUNCT fis-4216 148 29 1 1 NUM fis-4216 148 30 - - PUNCT fis-4216 148 31 β β X fis-4216 148 32 ) ) PUNCT fis-4216 148 33 e(θ e(θ ADJ fis-4216 148 34 ) ) PUNCT fis-4216 148 35 ( ( PUNCT fis-4216 148 36 4 4 X fis-4216 148 37 ) ) PUNCT fis-4216 148 38 ∇ ∇ X fis-4216 148 39 2 2 NUM fis-4216 148 40 l l NOUN fis-4216 148 41 2 2 NUM fis-4216 148 42 l-1 l-1 NOUN fis-4216 148 43 2 2 NUM fis-4216 148 44 lv lv PROPN fis-4216 148 45 = = NOUN fis-4216 148 46 β β NOUN fis-4216 148 47 v v ADP fis-4216 148 48 + + NOUN fis-4216 148 49 ( ( PUNCT fis-4216 148 50 1 1 NUM fis-4216 148 51 - - NOUN fis-4216 148 52 β β NOUN fis-4216 148 53 ) ) PUNCT fis-4216 148 54 [ [ PUNCT fis-4216 148 55 e(θ e(θ ADJ fis-4216 148 56 ) ) PUNCT fis-4216 148 57 ] ] PUNCT fis-4216 149 1 ( ( PUNCT fis-4216 149 2 5 5 X fis-4216 149 3 ) ) PUNCT fis-4216 149 4 where where SCONJ fis-4216 149 5 ℓ ℓ PROPN fis-4216 149 6 is be AUX fis-4216 149 7 the the DET fis-4216 149 8 iteration iteration NOUN fis-4216 149 9 number number NOUN fis-4216 149 10 , , PUNCT fis-4216 149 11 β1 β1 PROPN fis-4216 149 12 and and CCONJ fis-4216 149 13 β2 β2 NOUN fis-4216 149 14 are be AUX fis-4216 149 15 the the DET fis-4216 149 16 decay decay NOUN fis-4216 149 17 rates rate NOUN fis-4216 149 18 , , PUNCT fis-4216 149 19 θ θ PROPN fis-4216 149 20 is be AUX fis-4216 149 21 the the DET fis-4216 149 22 parameter parameter NOUN fis-4216 149 23 vector vector NOUN fis-4216 149 24 , , PUNCT fis-4216 149 25 and and CCONJ fis-4216 149 26 e(θ e(θ ADJ fis-4216 149 27 ) ) PUNCT fis-4216 149 28 is be AUX fis-4216 149 29 the the DET fis-4216 149 30 loss loss NOUN fis-4216 149 31 function function NOUN fis-4216 149 32 . . PUNCT fis-4216 150 1 the the DET fis-4216 150 2 adam adam PROPN fis-4216 150 3 algorithm algorithm PROPN fis-4216 150 4 updates update VERB fis-4216 150 5 the the DET fis-4216 150 6 network network NOUN fis-4216 150 7 parameters parameter NOUN fis-4216 150 8 using use VERB fis-4216 150 9 the the DET fis-4216 150 10 moving move VERB fis-4216 150 11 averages average NOUN fis-4216 150 12 as as ADP fis-4216 150 13 l. l. PROPN fis-4216 150 14 wang wang PROPN fis-4216 150 15 , , PUNCT fis-4216 150 16 frattura frattura PROPN fis-4216 150 17 ed ed PROPN fis-4216 150 18 integrità integrità PROPN fis-4216 150 19 strutturale strutturale PROPN fis-4216 150 20 , , PUNCT fis-4216 150 21 65 65 NUM fis-4216 150 22 ( ( PUNCT fis-4216 150 23 2023 2023 NUM fis-4216 150 24 ) ) PUNCT fis-4216 150 25 289 289 NUM fis-4216 150 26 - - SYM fis-4216 150 27 299 299 NUM fis-4216 150 28 ; ; PUNCT fis-4216 150 29 doi doi NOUN fis-4216 150 30 : : PUNCT fis-4216 150 31 10.3221 10.3221 NUM fis-4216 150 32 / / SYM fis-4216 150 33 igf igf PROPN fis-4216 150 34 - - PUNCT fis-4216 150 35 esis.65.19 esis.65.19 NOUN fis-4216 150 36 295 295 NUM fis-4216 150 37 l l NOUN fis-4216 150 38 l+1 l+1 X fis-4216 150 39 l l NOUN fis-4216 150 40 l l NOUN fis-4216 150 41 αm αm NOUN fis-4216 150 42 θ θ X fis-4216 151 1 = = PUNCT fis-4216 151 2 θ θ NOUN fis-4216 151 3     SPACE fis-4216 151 4 v v ADP fis-4216 151 5 + + PROPN fis-4216 151 6 ε ε PROPN fis-4216 151 7 ( ( PUNCT fis-4216 151 8 6 6 NUM fis-4216 151 9 ) ) PUNCT fis-4216 151 10 where where SCONJ fis-4216 151 11 α>0 α>0 NOUN fis-4216 151 12 is be AUX fis-4216 151 13 the the DET fis-4216 151 14 learning learning NOUN fis-4216 151 15 rate rate NOUN fis-4216 151 16 , , PUNCT fis-4216 151 17 and and CCONJ fis-4216 151 18 ε ε PROPN fis-4216 151 19 is be AUX fis-4216 151 20 a a DET fis-4216 151 21 small small ADJ fis-4216 151 22 positive positive ADJ fis-4216 151 23 value value NOUN fis-4216 151 24 . . PUNCT fis-4216 152 1 if if SCONJ fis-4216 152 2 the the DET fis-4216 152 3 gradients gradient NOUN fis-4216 152 4 are be AUX fis-4216 152 5 similar similar ADJ fis-4216 152 6 over over ADP fis-4216 152 7 many many ADJ fis-4216 152 8 iterations iteration NOUN fis-4216 152 9 , , PUNCT fis-4216 152 10 a a DET fis-4216 152 11 moving move VERB fis-4216 152 12 average average NOUN fis-4216 152 13 of of ADP fis-4216 152 14 the the DET fis-4216 152 15 gradient gradient NOUN fis-4216 152 16 allows allow VERB fis-4216 152 17 the the DET fis-4216 152 18 parameter parameter NOUN fis-4216 152 19 updates update VERB fis-4216 152 20 to to PART fis-4216 152 21 pick pick VERB fis-4216 152 22 up up ADP fis-4216 152 23 momentum momentum NOUN fis-4216 152 24 in in ADP fis-4216 152 25 a a DET fis-4216 152 26 particular particular ADJ fis-4216 152 27 direction direction NOUN fis-4216 152 28 . . PUNCT fis-4216 153 1 if if SCONJ fis-4216 153 2 the the DET fis-4216 153 3 gradients gradient NOUN fis-4216 153 4 contain contain VERB fis-4216 153 5 mostly mostly ADV fis-4216 153 6 noise noise NOUN fis-4216 153 7 , , PUNCT fis-4216 153 8 then then ADV fis-4216 153 9 the the DET fis-4216 153 10 moving move VERB fis-4216 153 11 average average NOUN fis-4216 153 12 of of ADP fis-4216 153 13 the the DET fis-4216 153 14 gradient gradient NOUN fis-4216 153 15 becomes become VERB fis-4216 153 16 smaller small ADJ fis-4216 153 17 , , PUNCT fis-4216 153 18 and and CCONJ fis-4216 153 19 so so ADV fis-4216 153 20 the the DET fis-4216 153 21 parameter parameter NOUN fis-4216 153 22 updates update NOUN fis-4216 153 23 become become VERB fis-4216 153 24 smaller small ADJ fis-4216 153 25 too too ADV fis-4216 153 26 . . PUNCT fis-4216 154 1 the the DET fis-4216 154 2 full full ADJ fis-4216 154 3 adam adam PROPN fis-4216 154 4 update update NOUN fis-4216 154 5 also also ADV fis-4216 154 6 includes include VERB fis-4216 154 7 a a DET fis-4216 154 8 mechanism mechanism NOUN fis-4216 154 9 to to PART fis-4216 154 10 correct correct VERB fis-4216 154 11 for for ADP fis-4216 154 12 any any DET fis-4216 154 13 bias bias NOUN fis-4216 154 14 that that PRON fis-4216 154 15 appears appear VERB fis-4216 154 16 at at ADP fis-4216 154 17 the the DET fis-4216 154 18 beginning beginning NOUN fis-4216 154 19 of of ADP fis-4216 154 20 training training NOUN fis-4216 154 21 . . PUNCT fis-4216 155 1 the the DET fis-4216 155 2 trained train VERB fis-4216 155 3 model model NOUN fis-4216 155 4 with with ADP fis-4216 155 5 optimized optimize VERB fis-4216 155 6 parameters parameter NOUN fis-4216 155 7 was be AUX fis-4216 155 8 later later ADV fis-4216 155 9 applied apply VERB fis-4216 155 10 to to ADP fis-4216 155 11 the the DET fis-4216 155 12 test test NOUN fis-4216 155 13 dataset dataset VERB fis-4216 155 14 for for ADP fis-4216 155 15 crack crack NOUN fis-4216 155 16 detection detection NOUN fis-4216 155 17 decision decision NOUN fis-4216 155 18 - - PUNCT fis-4216 155 19 making making NOUN fis-4216 155 20 . . PUNCT fis-4216 156 1 results result NOUN fis-4216 156 2 and and CCONJ fis-4216 156 3 discussion discussion NOUN fis-4216 156 4 utomatic utomatic ADJ fis-4216 156 5 crack crack NOUN fis-4216 156 6 detection detection NOUN fis-4216 156 7 system system NOUN fis-4216 156 8 cracksn cracksn NOUN fis-4216 156 9 was be AUX fis-4216 156 10 implemented implement VERB fis-4216 156 11 and and CCONJ fis-4216 156 12 tested test VERB fis-4216 156 13 in in ADP fis-4216 156 14 the the DET fis-4216 156 15 matlab matlab PROPN fis-4216 156 16 r2022b r2022b PROPN fis-4216 156 17 platform platform NOUN fis-4216 156 18 on on ADP fis-4216 156 19 a a DET fis-4216 156 20 desktop desktop NOUN fis-4216 156 21 with with ADP fis-4216 156 22 an an DET fis-4216 156 23 intel(r intel(r NOUN fis-4216 156 24 ) ) PUNCT fis-4216 156 25 core(tm core(tm NOUN fis-4216 156 26 ) ) PUNCT fis-4216 156 27 i7 i7 ADJ fis-4216 156 28 - - PUNCT fis-4216 156 29 13700kf 13700kf NOUN fis-4216 156 30 cpu cpu NOUN fis-4216 156 31 running run VERB fis-4216 156 32 at at ADP fis-4216 156 33 3.40ghz 3.40ghz NUM fis-4216 156 34 , , PUNCT fis-4216 156 35 64 64 NUM fis-4216 156 36 gb gb NOUN fis-4216 156 37 of of ADP fis-4216 156 38 ram ram NOUN fis-4216 156 39 , , PUNCT fis-4216 156 40 and and CCONJ fis-4216 156 41 an an DET fis-4216 156 42 nvidia nvidia PROPN fis-4216 156 43 rtx3090 rtx3090 X fis-4216 156 44 gpu gpu PROPN fis-4216 156 45 of of ADP fis-4216 156 46 24 24 NUM fis-4216 156 47 gb gb NOUN fis-4216 156 48 . . PUNCT fis-4216 157 1 the the DET fis-4216 157 2 network network NOUN fis-4216 157 3 was be AUX fis-4216 157 4 trained train VERB fis-4216 157 5 using use VERB fis-4216 157 6 the the DET fis-4216 157 7 adam adam PROPN fis-4216 157 8 optimizer optimizer NOUN fis-4216 157 9 with with ADP fis-4216 157 10 a a DET fis-4216 157 11 batch batch NOUN fis-4216 157 12 size size NOUN fis-4216 157 13 of of ADP fis-4216 157 14 100 100 NUM fis-4216 157 15 examples example NOUN fis-4216 157 16 and and CCONJ fis-4216 157 17 an an DET fis-4216 157 18 initial initial ADJ fis-4216 157 19 learning learning NOUN fis-4216 157 20 rate rate NOUN fis-4216 157 21 of of ADP fis-4216 157 22 0.0002 0.0002 NUM fis-4216 157 23 . . PUNCT fis-4216 158 1 the the DET fis-4216 158 2 performance performance NOUN fis-4216 158 3 of of ADP fis-4216 158 4 the the DET fis-4216 158 5 implemented implement VERB fis-4216 158 6 system system NOUN fis-4216 158 7 was be AUX fis-4216 158 8 evaluated evaluate VERB fis-4216 158 9 using use VERB fis-4216 158 10 three three NUM fis-4216 158 11 metrics metric NOUN fis-4216 158 12 . . PUNCT fis-4216 159 1 recall recall NOUN fis-4216 159 2 ( ( PUNCT fis-4216 159 3 rec rec NOUN fis-4216 159 4 ) ) PUNCT fis-4216 159 5 , , PUNCT fis-4216 159 6 or or CCONJ fis-4216 159 7 accuracy accuracy NOUN fis-4216 159 8 , , PUNCT fis-4216 159 9 is be AUX fis-4216 159 10 the the DET fis-4216 159 11 ratio ratio NOUN fis-4216 159 12 of of ADP fis-4216 159 13 correct correct ADJ fis-4216 159 14 crack crack NOUN fis-4216 159 15 predictions prediction NOUN fis-4216 159 16 to to ADP fis-4216 159 17 the the DET fis-4216 159 18 number number NOUN fis-4216 159 19 of of ADP fis-4216 159 20 crack crack NOUN fis-4216 159 21 patches patch NOUN fis-4216 159 22 . . PUNCT fis-4216 160 1 precision precision NOUN fis-4216 160 2 ( ( PUNCT fis-4216 160 3 pre pre ADJ fis-4216 160 4 ) ) PUNCT fis-4216 160 5 is be AUX fis-4216 160 6 the the DET fis-4216 160 7 ratio ratio NOUN fis-4216 160 8 of of ADP fis-4216 160 9 correct correct ADJ fis-4216 160 10 crack crack NOUN fis-4216 160 11 precisions precision NOUN fis-4216 160 12 to to PART fis-4216 160 13 total total ADJ fis-4216 160 14 crack crack NOUN fis-4216 160 15 predictions prediction NOUN fis-4216 160 16 . . PUNCT fis-4216 161 1 the the DET fis-4216 161 2 f1 f1 PROPN fis-4216 161 3 score score NOUN fis-4216 161 4 is be AUX fis-4216 161 5 the the DET fis-4216 161 6 harmonic harmonic ADJ fis-4216 161 7 mean mean NOUN fis-4216 161 8 of of ADP fis-4216 161 9 precision precision NOUN fis-4216 161 10 and and CCONJ fis-4216 161 11 recall recall NOUN fis-4216 161 12 . . PUNCT fis-4216 162 1 the the DET fis-4216 162 2 cracked crack VERB fis-4216 162 3 and and CCONJ fis-4216 162 4 uncracked uncracked ADJ fis-4216 162 5 patches patch NOUN fis-4216 162 6 were be AUX fis-4216 162 7 designated designate VERB fis-4216 162 8 as as ADP fis-4216 162 9 positive positive ADJ fis-4216 162 10 and and CCONJ fis-4216 162 11 negative negative ADJ fis-4216 162 12 , , PUNCT fis-4216 162 13 accordingly accordingly ADV fis-4216 162 14 . . PUNCT fis-4216 163 1 tp tp ADP fis-4216 163 2 rec= rec= PROPN fis-4216 163 3 tp+fn tp+fn NUM fis-4216 163 4 ( ( PUNCT fis-4216 163 5 7 7 NUM fis-4216 163 6 ) ) PUNCT fis-4216 163 7 tp tp PART fis-4216 163 8 pre= pre= VERB fis-4216 163 9 tp+fp tp+fp NUM fis-4216 163 10 ( ( PUNCT fis-4216 163 11 8) 8) NUM fis-4216 163 12 ⋅ ⋅ X fis-4216 163 13 ⋅2 ⋅2 PROPN fis-4216 163 14 pre pre X fis-4216 163 15 rec rec PROPN fis-4216 163 16 f1= f1= PROPN fis-4216 163 17 pre+rec pre+rec PROPN fis-4216 163 18 ( ( PUNCT fis-4216 163 19 9 9 NUM fis-4216 163 20 ) ) PUNCT fis-4216 163 21 where where SCONJ fis-4216 163 22 , , PUNCT fis-4216 163 23 in in ADP fis-4216 163 24 turn turn NOUN fis-4216 163 25 , , PUNCT fis-4216 163 26 tp tp NOUN fis-4216 163 27 , , PUNCT fis-4216 163 28 tn tn PROPN fis-4216 163 29 , , PUNCT fis-4216 163 30 fp fp NOUN fis-4216 163 31 , , PUNCT fis-4216 163 32 and and CCONJ fis-4216 163 33 fn fn NOUN fis-4216 163 34 stand stand VERB fis-4216 163 35 for for ADP fis-4216 163 36 the the DET fis-4216 163 37 corresponding corresponding ADJ fis-4216 163 38 true true ADJ fis-4216 163 39 positive positive ADJ fis-4216 163 40 , , PUNCT fis-4216 163 41 true true ADJ fis-4216 163 42 negative negative ADJ fis-4216 163 43 , , PUNCT fis-4216 163 44 false false ADJ fis-4216 163 45 positive positive ADJ fis-4216 163 46 , , PUNCT fis-4216 163 47 and and CCONJ fis-4216 163 48 false false ADJ fis-4216 163 49 negative negative ADJ fis-4216 163 50 . . PUNCT fis-4216 164 1 fig fig NOUN fis-4216 164 2 . . PUNCT fis-4216 165 1 4 4 NUM fis-4216 165 2 depicts depict VERB fis-4216 165 3 the the DET fis-4216 165 4 changes change NOUN fis-4216 165 5 in in ADP fis-4216 165 6 loss loss NOUN fis-4216 165 7 function function NOUN fis-4216 165 8 and and CCONJ fis-4216 165 9 crack crack VERB fis-4216 165 10 accuracy accuracy NOUN fis-4216 165 11 of of ADP fis-4216 165 12 the the DET fis-4216 165 13 squeeznet squeeznet NOUN fis-4216 165 14 - - PUNCT fis-4216 165 15 based base VERB fis-4216 165 16 model model NOUN fis-4216 165 17 during during ADP fis-4216 165 18 its its PRON fis-4216 165 19 training training NOUN fis-4216 165 20 and and CCONJ fis-4216 165 21 validation validation NOUN fis-4216 165 22 processes process NOUN fis-4216 165 23 . . PUNCT fis-4216 166 1 the the DET fis-4216 166 2 loss loss NOUN fis-4216 166 3 function function NOUN fis-4216 166 4 achieves achieve VERB fis-4216 166 5 its its PRON fis-4216 166 6 minimum minimum NOUN fis-4216 166 7 after after ADP fis-4216 166 8 ten ten NUM fis-4216 166 9 epochs epoch NOUN fis-4216 166 10 on on ADP fis-4216 166 11 the the DET fis-4216 166 12 training training NOUN fis-4216 166 13 set set NOUN fis-4216 166 14 , , PUNCT fis-4216 166 15 indicating indicate VERB fis-4216 166 16 the the DET fis-4216 166 17 convergence convergence NOUN fis-4216 166 18 of of ADP fis-4216 166 19 the the DET fis-4216 166 20 weights weight NOUN fis-4216 166 21 . . PUNCT fis-4216 167 1 the the DET fis-4216 167 2 detailed detailed ADJ fis-4216 167 3 performance performance NOUN fis-4216 167 4 of of ADP fis-4216 167 5 the the DET fis-4216 167 6 implemented implement VERB fis-4216 167 7 crack crack NOUN fis-4216 167 8 detection detection NOUN fis-4216 167 9 system system NOUN fis-4216 167 10 , , PUNCT fis-4216 167 11 cracksn cracksn NOUN fis-4216 167 12 , , PUNCT fis-4216 167 13 on on ADP fis-4216 167 14 the the DET fis-4216 167 15 training training NOUN fis-4216 167 16 and and CCONJ fis-4216 167 17 validation validation NOUN fis-4216 167 18 image image NOUN fis-4216 167 19 datasets dataset NOUN fis-4216 167 20 , , PUNCT fis-4216 167 21 is be AUX fis-4216 167 22 presented present VERB fis-4216 167 23 in in ADP fis-4216 167 24 tab tab NOUN fis-4216 167 25 . . PUNCT fis-4216 168 1 2 2 X fis-4216 168 2 . . X fis-4216 168 3 the the DET fis-4216 168 4 implemented implement VERB fis-4216 168 5 squeezenet squeezenet NOUN fis-4216 168 6 - - PUNCT fis-4216 168 7 based base VERB fis-4216 168 8 model model NOUN fis-4216 168 9 was be AUX fis-4216 168 10 compared compare VERB fis-4216 168 11 with with ADP fis-4216 168 12 state state NOUN fis-4216 168 13 - - PUNCT fis-4216 168 14 ofthe ofthe NOUN fis-4216 168 15 - - PUNCT fis-4216 168 16 art art NOUN fis-4216 168 17 models model NOUN fis-4216 168 18 such such ADJ fis-4216 168 19 as as ADP fis-4216 168 20 crackpix crackpix ADJ fis-4216 168 21 fcn2s fcn2s PUNCT fis-4216 169 1 [ [ X fis-4216 169 2 4 4 NUM fis-4216 169 3 ] ] PUNCT fis-4216 169 4 , , PUNCT fis-4216 169 5 classification classification NOUN fis-4216 169 6 vgg16 vgg16 NOUN fis-4216 170 1 [ [ NOUN fis-4216 170 2 4 4 NUM fis-4216 170 3 ] ] PUNCT fis-4216 170 4 , , PUNCT fis-4216 170 5 and and CCONJ fis-4216 170 6 convnet convnet NOUN fis-4216 171 1 [ [ X fis-4216 171 2 11 11 NUM fis-4216 171 3 ] ] PUNCT fis-4216 171 4 . . PUNCT fis-4216 172 1 it it PRON fis-4216 172 2 should should AUX fis-4216 172 3 be be AUX fis-4216 172 4 noted note VERB fis-4216 172 5 that that SCONJ fis-4216 172 6 every every DET fis-4216 172 7 model model NOUN fis-4216 172 8 was be AUX fis-4216 172 9 trained train VERB fis-4216 172 10 on on ADP fis-4216 172 11 different different ADJ fis-4216 172 12 image image NOUN fis-4216 172 13 datasets dataset NOUN fis-4216 172 14 . . PUNCT fis-4216 173 1 in in ADP fis-4216 173 2 addition addition NOUN fis-4216 173 3 , , PUNCT fis-4216 173 4 the the DET fis-4216 173 5 crackpix crackpix ADJ fis-4216 173 6 fcn2s fcn2s PROPN fis-4216 173 7 model model NOUN fis-4216 173 8 provides provide VERB fis-4216 173 9 crack crack NOUN fis-4216 173 10 identification identification NOUN fis-4216 173 11 at at ADP fis-4216 173 12 the the DET fis-4216 173 13 pixel pixel PROPN fis-4216 173 14 level level NOUN fis-4216 173 15 , , PUNCT fis-4216 173 16 as as SCONJ fis-4216 173 17 opposed oppose VERB fis-4216 173 18 to to ADP fis-4216 173 19 the the DET fis-4216 173 20 patch patch ADJ fis-4216 173 21 level level NOUN fis-4216 173 22 of of ADP fis-4216 173 23 the the DET fis-4216 173 24 other other ADJ fis-4216 173 25 models model NOUN fis-4216 173 26 . . PUNCT fis-4216 174 1 compared compare VERB fis-4216 174 2 to to ADP fis-4216 174 3 the the DET fis-4216 174 4 reported report VERB fis-4216 174 5 results result NOUN fis-4216 174 6 of of ADP fis-4216 174 7 the the DET fis-4216 174 8 models model NOUN fis-4216 174 9 mentioned mention VERB fis-4216 174 10 above above ADV fis-4216 174 11 in in ADP fis-4216 174 12 the the DET fis-4216 174 13 literature literature NOUN fis-4216 174 14 , , PUNCT fis-4216 174 15 our our PRON fis-4216 174 16 model model NOUN fis-4216 174 17 provides provide VERB fis-4216 174 18 an an DET fis-4216 174 19 encouraging encouraging ADJ fis-4216 174 20 performance performance NOUN fis-4216 174 21 in in ADP fis-4216 174 22 detecting detect VERB fis-4216 174 23 the the DET fis-4216 174 24 concrete concrete ADJ fis-4216 174 25 crack crack NOUN fis-4216 174 26 at at ADP fis-4216 174 27 the the DET fis-4216 174 28 patch patch ADJ fis-4216 174 29 level level NOUN fis-4216 174 30 . . PUNCT fis-4216 175 1 the the DET fis-4216 175 2 values value NOUN fis-4216 175 3 of of ADP fis-4216 175 4 rec rec NOUN fis-4216 175 5 , , PUNCT fis-4216 175 6 pre pre ADJ fis-4216 175 7 , , PUNCT fis-4216 175 8 and and CCONJ fis-4216 175 9 f1 f1 NOUN fis-4216 175 10 metrics metric NOUN fis-4216 175 11 demonstrate demonstrate VERB fis-4216 175 12 the the DET fis-4216 175 13 effectiveness effectiveness NOUN fis-4216 175 14 of of ADP fis-4216 175 15 our our PRON fis-4216 175 16 model model NOUN fis-4216 175 17 cracksn cracksn NOUN fis-4216 175 18 for for ADP fis-4216 175 19 automatic automatic ADJ fis-4216 175 20 crack crack NOUN fis-4216 175 21 detection detection NOUN fis-4216 175 22 from from ADP fis-4216 175 23 concrete concrete ADJ fis-4216 175 24 surface surface NOUN fis-4216 175 25 images image NOUN fis-4216 175 26 . . PUNCT fis-4216 176 1 meanwhile meanwhile ADV fis-4216 176 2 , , PUNCT fis-4216 176 3 the the DET fis-4216 176 4 presented present VERB fis-4216 176 5 system system NOUN fis-4216 176 6 , , PUNCT fis-4216 176 7 cracksn cracksn NOUN fis-4216 176 8 , , PUNCT fis-4216 176 9 based base VERB fis-4216 176 10 on on ADP fis-4216 176 11 the the DET fis-4216 176 12 adam adam NOUN fis-4216 176 13 - - PUNCT fis-4216 176 14 squeezenet squeezenet NOUN fis-4216 176 15 deep deep ADJ fis-4216 176 16 learning learning NOUN fis-4216 176 17 architecture architecture NOUN fis-4216 176 18 , , PUNCT fis-4216 176 19 has have VERB fis-4216 176 20 only only ADV fis-4216 176 21 1.24 1.24 NUM fis-4216 176 22 million million NUM fis-4216 176 23 parameters parameter NOUN fis-4216 176 24 and and CCONJ fis-4216 176 25 a a DET fis-4216 176 26 model model NOUN fis-4216 176 27 size size NOUN fis-4216 176 28 ( ( PUNCT fis-4216 176 29 implemented implement VERB fis-4216 176 30 in in ADP fis-4216 176 31 matlab matlab PROPN fis-4216 176 32 r2022b r2022b NOUN fis-4216 176 33 ) ) PUNCT fis-4216 176 34 of of ADP fis-4216 176 35 around around ADV fis-4216 176 36 5 5 NUM fis-4216 176 37 mb mb NOUN fis-4216 176 38 . . PUNCT fis-4216 177 1 although although SCONJ fis-4216 177 2 a a DET fis-4216 177 3 direct direct ADJ fis-4216 177 4 efficiency efficiency NOUN fis-4216 177 5 comparison comparison NOUN fis-4216 177 6 of of ADP fis-4216 177 7 the the DET fis-4216 177 8 convnet convnet NOUN fis-4216 177 9 , , PUNCT fis-4216 177 10 the the DET fis-4216 177 11 classification classification NOUN fis-4216 177 12 vgg16 vgg16 NOUN fis-4216 177 13 , , PUNCT fis-4216 177 14 and and CCONJ fis-4216 177 15 the the DET fis-4216 177 16 crackpix crackpix ADJ fis-4216 177 17 fcn2s fcn2s PROPN fis-4216 177 18 models model NOUN fis-4216 177 19 is be AUX fis-4216 177 20 not not PART fis-4216 177 21 available available ADJ fis-4216 177 22 for for ADP fis-4216 177 23 testing test VERB fis-4216 177 24 on on ADP fis-4216 177 25 the the DET fis-4216 177 26 same same ADJ fis-4216 177 27 image image NOUN fis-4216 177 28 dataset dataset NOUN fis-4216 177 29 , , PUNCT fis-4216 177 30 the the DET fis-4216 177 31 training training NOUN fis-4216 177 32 expense expense NOUN fis-4216 177 33 of of ADP fis-4216 177 34 these these DET fis-4216 177 35 models model NOUN fis-4216 177 36 can can AUX fis-4216 177 37 be be AUX fis-4216 177 38 indirectly indirectly ADV fis-4216 177 39 evaluated evaluate VERB fis-4216 177 40 from from ADP fis-4216 177 41 their their PRON fis-4216 177 42 backbone backbone NOUN fis-4216 177 43 networks network NOUN fis-4216 177 44 . . PUNCT fis-4216 178 1 the the DET fis-4216 178 2 convnet convnet PROPN fis-4216 178 3 model model NOUN fis-4216 178 4 built build VERB fis-4216 178 5 on on ADP fis-4216 178 6 the the DET fis-4216 178 7 cnn cnn PROPN fis-4216 178 8 network network NOUN fis-4216 178 9 has have VERB fis-4216 178 10 more more ADJ fis-4216 178 11 than than ADP fis-4216 178 12 60 60 NUM fis-4216 178 13 million million NUM fis-4216 178 14 parameters parameter NOUN fis-4216 178 15 and and CCONJ fis-4216 178 16 a a DET fis-4216 178 17 code code NOUN fis-4216 178 18 size size NOUN fis-4216 178 19 of of ADP fis-4216 178 20 227 227 NUM fis-4216 178 21 mb mb NOUN fis-4216 178 22 . . PUNCT fis-4216 179 1 the the DET fis-4216 179 2 classification classification NOUN fis-4216 179 3 vgg16 vgg16 NOUN fis-4216 179 4 and and CCONJ fis-4216 179 5 the the DET fis-4216 179 6 crackpix crackpix ADJ fis-4216 179 7 fcn2s fcn2s NOUN fis-4216 179 8 adopt adopt VERB fis-4216 179 9 the the DET fis-4216 179 10 vgg16 vgg16 PROPN fis-4216 179 11 and and CCONJ fis-4216 179 12 vgg19 vgg19 PROPN fis-4216 179 13 backbones backbone NOUN fis-4216 179 14 , , PUNCT fis-4216 179 15 which which PRON fis-4216 179 16 have have VERB fis-4216 179 17 about about ADP fis-4216 179 18 140 140 NUM fis-4216 179 19 million million NUM fis-4216 179 20 parameters parameter NOUN fis-4216 179 21 and and CCONJ fis-4216 179 22 more more ADJ fis-4216 179 23 than than ADP fis-4216 179 24 500 500 NUM fis-4216 179 25 mb mb NOUN fis-4216 179 26 in in ADP fis-4216 179 27 code code NOUN fis-4216 179 28 implementation implementation NOUN fis-4216 179 29 . . PUNCT fis-4216 180 1 the the DET fis-4216 180 2 deep deep ADJ fis-4216 180 3 learning learning NOUN fis-4216 180 4 structure structure NOUN fis-4216 180 5 of of ADP fis-4216 180 6 the the DET fis-4216 180 7 proposed propose VERB fis-4216 180 8 system system NOUN fis-4216 180 9 in in ADP fis-4216 180 10 this this DET fis-4216 180 11 work work NOUN fis-4216 180 12 is be AUX fis-4216 180 13 much much ADV fis-4216 180 14 more more ADV fis-4216 180 15 compact compact ADJ fis-4216 180 16 than than ADP fis-4216 180 17 the the DET fis-4216 180 18 other other ADJ fis-4216 180 19 models model NOUN fis-4216 180 20 listed list VERB fis-4216 180 21 in in ADP fis-4216 180 22 tab tab NOUN fis-4216 180 23 . . PUNCT fis-4216 181 1 2 2 X fis-4216 181 2 . . X fis-4216 181 3 when when SCONJ fis-4216 181 4 integrated integrate VERB fis-4216 181 5 with with ADP fis-4216 181 6 cloud cloud NOUN fis-4216 181 7 storage storage NOUN fis-4216 181 8 and and CCONJ fis-4216 181 9 an an DET fis-4216 181 10 unmanned unmanned ADJ fis-4216 181 11 aerial aerial ADJ fis-4216 181 12 vehicle vehicle NOUN fis-4216 181 13 for for ADP fis-4216 181 14 image image NOUN fis-4216 181 15 acquisition acquisition NOUN fis-4216 181 16 , , PUNCT fis-4216 181 17 the the DET fis-4216 181 18 adam adam NOUN fis-4216 181 19 - - PUNCT fis-4216 181 20 squeezenet squeezenet NOUN fis-4216 181 21 - - PUNCT fis-4216 181 22 based base VERB fis-4216 181 23 system system NOUN fis-4216 181 24 cracksn cracksn NOUN fis-4216 181 25 can can AUX fis-4216 181 26 potentially potentially ADV fis-4216 181 27 be be AUX fis-4216 181 28 used use VERB fis-4216 181 29 for for ADP fis-4216 181 30 real real ADJ fis-4216 181 31 - - PUNCT fis-4216 181 32 time time NOUN fis-4216 181 33 and and CCONJ fis-4216 181 34 remote remote ADJ fis-4216 181 35 health health NOUN fis-4216 181 36 diagnosis diagnosis NOUN fis-4216 181 37 of of ADP fis-4216 181 38 concrete concrete ADJ fis-4216 181 39 infrastructure infrastructure NOUN fis-4216 181 40 . . PUNCT fis-4216 182 1 the the DET fis-4216 182 2 advantage advantage NOUN fis-4216 182 3 in in ADP fis-4216 182 4 computational computational ADJ fis-4216 182 5 efficiency efficiency NOUN fis-4216 182 6 and and CCONJ fis-4216 182 7 lightweight lightweight ADJ fis-4216 182 8 structure structure NOUN fis-4216 182 9 of of ADP fis-4216 182 10 the the DET fis-4216 182 11 presented present VERB fis-4216 182 12 system system NOUN fis-4216 182 13 enables enable VERB fis-4216 182 14 the the DET fis-4216 182 15 work work NOUN fis-4216 182 16 to to PART fis-4216 182 17 be be AUX fis-4216 182 18 mobile mobile ADJ fis-4216 182 19 , , PUNCT fis-4216 182 20 appealing appeal VERB fis-4216 182 21 to to ADP fis-4216 182 22 civil civil ADJ fis-4216 182 23 engineers engineer NOUN fis-4216 182 24 for for ADP fis-4216 182 25 structure structure NOUN fis-4216 182 26 damage damage NOUN fis-4216 182 27 evaluation evaluation NOUN fis-4216 182 28 with with ADP fis-4216 182 29 satisfied satisfied ADJ fis-4216 182 30 performance performance NOUN fis-4216 182 31 . . PUNCT fis-4216 183 1 a a DET fis-4216 183 2 l. l. PROPN fis-4216 183 3 wang wang PROPN fis-4216 183 4 , , PUNCT fis-4216 183 5 frattura frattura PROPN fis-4216 183 6 ed ed PROPN fis-4216 183 7 integrità integrità PROPN fis-4216 183 8 strutturale strutturale PROPN fis-4216 183 9 , , PUNCT fis-4216 183 10 65 65 NUM fis-4216 183 11 ( ( PUNCT fis-4216 183 12 2023 2023 NUM fis-4216 183 13 ) ) PUNCT fis-4216 183 14 289 289 NUM fis-4216 183 15 - - SYM fis-4216 183 16 299 299 NUM fis-4216 183 17 ; ; PUNCT fis-4216 183 18 doi doi NOUN fis-4216 183 19 : : PUNCT fis-4216 183 20 10.3221 10.3221 NUM fis-4216 183 21 / / SYM fis-4216 183 22 igf igf PROPN fis-4216 183 23 - - PUNCT fis-4216 183 24 esis.65.19 esis.65.19 NOUN fis-4216 183 25 296 296 NUM fis-4216 183 26 fig fig NOUN fis-4216 183 27 . . PUNCT fis-4216 183 28 5 5 NUM fis-4216 183 29 shows show VERB fis-4216 183 30 the the DET fis-4216 183 31 test test NOUN fis-4216 183 32 image image NOUN fis-4216 183 33 and and CCONJ fis-4216 183 34 the the DET fis-4216 183 35 calculated calculate VERB fis-4216 183 36 probability probability NOUN fis-4216 183 37 map map NOUN fis-4216 183 38 of of ADP fis-4216 183 39 correct correct ADJ fis-4216 183 40 classification classification NOUN fis-4216 183 41 by by ADP fis-4216 183 42 the the DET fis-4216 183 43 proposed propose VERB fis-4216 183 44 crack crack NOUN fis-4216 183 45 detection detection NOUN fis-4216 183 46 system system NOUN fis-4216 183 47 , , PUNCT fis-4216 183 48 cracksn cracksn NOUN fis-4216 183 49 . . PUNCT fis-4216 184 1 the the DET fis-4216 184 2 probability probability NOUN fis-4216 184 3 map map NOUN fis-4216 184 4 of of ADP fis-4216 184 5 the the DET fis-4216 184 6 pixel pixel NOUN fis-4216 184 7 in in ADP fis-4216 184 8 red red ADJ fis-4216 184 9 indicates indicate VERB fis-4216 184 10 high high ADJ fis-4216 184 11 confidence confidence NOUN fis-4216 184 12 in in ADP fis-4216 184 13 classifying classify VERB fis-4216 184 14 the the DET fis-4216 184 15 image image NOUN fis-4216 184 16 as as SCONJ fis-4216 184 17 cracked crack VERB fis-4216 184 18 or or CCONJ fis-4216 184 19 not not PART fis-4216 184 20 . . PUNCT fis-4216 185 1 for for ADP fis-4216 185 2 instance instance NOUN fis-4216 185 3 , , PUNCT fis-4216 185 4 normal normal ADJ fis-4216 185 5 images image NOUN fis-4216 185 6 without without ADP fis-4216 185 7 crack crack NOUN fis-4216 185 8 characteristics characteristic NOUN fis-4216 185 9 typically typically ADV fis-4216 185 10 have have VERB fis-4216 185 11 a a DET fis-4216 185 12 large large ADJ fis-4216 185 13 , , PUNCT fis-4216 185 14 connected connected ADJ fis-4216 185 15 area area NOUN fis-4216 185 16 with with ADP fis-4216 185 17 high high ADJ fis-4216 185 18 confidence confidence NOUN fis-4216 185 19 . . PUNCT fis-4216 186 1 in in ADP fis-4216 186 2 contrast contrast NOUN fis-4216 186 3 to to ADP fis-4216 186 4 normal normal ADJ fis-4216 186 5 or or CCONJ fis-4216 186 6 uncracked uncracked ADJ fis-4216 186 7 images image NOUN fis-4216 186 8 , , PUNCT fis-4216 186 9 the the DET fis-4216 186 10 probability probability NOUN fis-4216 186 11 maps map NOUN fis-4216 186 12 of of ADP fis-4216 186 13 cracked crack VERB fis-4216 186 14 images image NOUN fis-4216 186 15 are be AUX fis-4216 186 16 highly highly ADV fis-4216 186 17 localized localized ADJ fis-4216 186 18 to to ADP fis-4216 186 19 the the DET fis-4216 186 20 cracked crack VERB fis-4216 186 21 pixels pixel NOUN fis-4216 186 22 . . PUNCT fis-4216 187 1 these these DET fis-4216 187 2 results result NOUN fis-4216 187 3 demonstrate demonstrate VERB fis-4216 187 4 the the DET fis-4216 187 5 discriminative discriminative NOUN fis-4216 187 6 capability capability NOUN fis-4216 187 7 of of ADP fis-4216 187 8 the the DET fis-4216 187 9 presented present VERB fis-4216 187 10 deep deep ADJ fis-4216 187 11 learning learning NOUN fis-4216 187 12 model model NOUN fis-4216 187 13 . . PUNCT fis-4216 188 1 for for ADP fis-4216 188 2 further further ADJ fis-4216 188 3 evaluation evaluation NOUN fis-4216 188 4 , , PUNCT fis-4216 188 5 the the DET fis-4216 188 6 ground ground NOUN fis-4216 188 7 truth truth NOUN fis-4216 188 8 segmentation segmentation NOUN fis-4216 188 9 of of ADP fis-4216 188 10 concrete concrete ADJ fis-4216 188 11 cracks crack NOUN fis-4216 188 12 at at ADP fis-4216 188 13 the the DET fis-4216 188 14 pixel pixel PROPN fis-4216 188 15 level level NOUN fis-4216 188 16 was be AUX fis-4216 188 17 obtained obtain VERB fis-4216 188 18 from from ADP fis-4216 188 19 the the DET fis-4216 188 20 identified identify VERB fis-4216 188 21 positive positive ADJ fis-4216 188 22 or or CCONJ fis-4216 188 23 cracked crack VERB fis-4216 188 24 patches patch NOUN fis-4216 188 25 and and CCONJ fis-4216 188 26 shown show VERB fis-4216 188 27 in in ADP fis-4216 188 28 fig fig NOUN fis-4216 188 29 . . PUNCT fis-4216 189 1 6 6 X fis-4216 189 2 . . X fis-4216 189 3 crack crack NOUN fis-4216 189 4 size size NOUN fis-4216 189 5 is be AUX fis-4216 189 6 an an DET fis-4216 189 7 essential essential ADJ fis-4216 189 8 reference reference NOUN fis-4216 189 9 when when SCONJ fis-4216 189 10 assessing assess VERB fis-4216 189 11 the the DET fis-4216 189 12 in in ADP fis-4216 189 13 - - PUNCT fis-4216 189 14 service service NOUN fis-4216 189 15 condition condition NOUN fis-4216 189 16 of of ADP fis-4216 189 17 concrete concrete ADJ fis-4216 189 18 buildings building NOUN fis-4216 189 19 [ [ X fis-4216 189 20 3 3 NUM fis-4216 189 21 - - SYM fis-4216 189 22 5 5 NUM fis-4216 189 23 , , PUNCT fis-4216 189 24 14 14 NUM fis-4216 189 25 ] ] PUNCT fis-4216 189 26 . . PUNCT fis-4216 190 1 the the DET fis-4216 190 2 crack crack NOUN fis-4216 190 3 size size NOUN fis-4216 190 4 information information NOUN fis-4216 190 5 , , PUNCT fis-4216 190 6 such such ADJ fis-4216 190 7 as as ADP fis-4216 190 8 the the DET fis-4216 190 9 type type NOUN fis-4216 190 10 , , PUNCT fis-4216 190 11 width width ADJ fis-4216 190 12 , , PUNCT fis-4216 190 13 length length NOUN fis-4216 190 14 , , PUNCT fis-4216 190 15 and and CCONJ fis-4216 190 16 location location NOUN fis-4216 190 17 of of ADP fis-4216 190 18 the the DET fis-4216 190 19 visible visible ADJ fis-4216 190 20 crack crack NOUN fis-4216 190 21 , , PUNCT fis-4216 190 22 is be AUX fis-4216 190 23 crucial crucial ADJ fis-4216 190 24 to to ADP fis-4216 190 25 evaluating evaluate VERB fis-4216 190 26 the the DET fis-4216 190 27 severity severity NOUN fis-4216 190 28 of of ADP fis-4216 190 29 cracking crack VERB fis-4216 190 30 development development NOUN fis-4216 190 31 and and CCONJ fis-4216 190 32 the the DET fis-4216 190 33 evolution evolution NOUN fis-4216 190 34 of of ADP fis-4216 190 35 damage damage NOUN fis-4216 190 36 in in ADP fis-4216 190 37 the the DET fis-4216 190 38 structure structure NOUN fis-4216 190 39 . . PUNCT fis-4216 191 1 decisions decision NOUN fis-4216 191 2 on on ADP fis-4216 191 3 subsequent subsequent ADJ fis-4216 191 4 maintenance maintenance NOUN fis-4216 191 5 could could AUX fis-4216 191 6 be be AUX fis-4216 191 7 made make VERB fis-4216 191 8 by by ADP fis-4216 191 9 tracking tracking NOUN fis-4216 191 10 and and CCONJ fis-4216 191 11 comparing compare VERB fis-4216 191 12 the the DET fis-4216 191 13 cracking crack VERB fis-4216 191 14 condition condition NOUN fis-4216 191 15 of of ADP fis-4216 191 16 a a DET fis-4216 191 17 concrete concrete ADJ fis-4216 191 18 infrastructure infrastructure NOUN fis-4216 191 19 at at ADP fis-4216 191 20 each each DET fis-4216 191 21 inspection inspection NOUN fis-4216 191 22 to to PART fis-4216 191 23 ensure ensure VERB fis-4216 191 24 the the DET fis-4216 191 25 serviceability serviceability NOUN fis-4216 191 26 and and CCONJ fis-4216 191 27 integrity integrity NOUN fis-4216 191 28 of of ADP fis-4216 191 29 the the DET fis-4216 191 30 structure structure NOUN fis-4216 191 31 . . PUNCT fis-4216 192 1 figure figure VERB fis-4216 192 2 4 4 NUM fis-4216 192 3 : : PUNCT fis-4216 192 4 development development NOUN fis-4216 192 5 of of ADP fis-4216 192 6 loss loss NOUN fis-4216 192 7 and and CCONJ fis-4216 192 8 crack crack VERB fis-4216 192 9 accuracy accuracy NOUN fis-4216 192 10 rec rec NOUN fis-4216 192 11 during during ADP fis-4216 192 12 training training NOUN fis-4216 192 13 of of ADP fis-4216 192 14 the the DET fis-4216 192 15 cracksn cracksn PROPN fis-4216 192 16 model model PROPN fis-4216 192 17 . . PUNCT fis-4216 193 1 model model PROPN fis-4216 193 2 rec rec PROPN fis-4216 193 3 pre pre PROPN fis-4216 193 4 f1 f1 PROPN fis-4216 193 5 cracksn cracksn NOUN fis-4216 193 6 0.973 0.973 NUM fis-4216 193 7 0.986 0.986 NUM fis-4216 193 8 0.962 0.962 NUM fis-4216 193 9 convnet[11 convnet[11 NOUN fis-4216 193 10 ] ] PUNCT fis-4216 193 11 0.870 0.870 NUM fis-4216 193 12 0.925 0.925 NUM fis-4216 193 13 0.897 0.897 NUM fis-4216 193 14 vgg16[4 vgg16[4 NOUN fis-4216 193 15 ] ] PUNCT fis-4216 193 16 0.938 0.938 NUM fis-4216 193 17 0.919 0.919 NUM fis-4216 193 18 0.928 0.928 NUM fis-4216 193 19 crackpix crackpix ADJ fis-4216 193 20 fcn2s[4 fcn2s[4 NOUN fis-4216 193 21 ] ] PUNCT fis-4216 193 22 0.922 0.922 NUM fis-4216 193 23 0.912 0.912 NUM fis-4216 193 24 0.917 0.917 NUM fis-4216 193 25 table table NOUN fis-4216 193 26 2 2 NUM fis-4216 193 27 : : PUNCT fis-4216 193 28 performance performance NOUN fis-4216 193 29 of of ADP fis-4216 193 30 patch patch ADJ fis-4216 193 31 - - PUNCT fis-4216 193 32 level level NOUN fis-4216 193 33 and and CCONJ fis-4216 193 34 pixel pixel ADJ fis-4216 193 35 - - PUNCT fis-4216 193 36 level level NOUN fis-4216 193 37 models model NOUN fis-4216 193 38 . . PUNCT fis-4216 194 1 the the DET fis-4216 194 2 experimental experimental ADJ fis-4216 194 3 results result NOUN fis-4216 194 4 of of ADP fis-4216 194 5 the the DET fis-4216 194 6 presented present VERB fis-4216 194 7 system system NOUN fis-4216 194 8 prove prove VERB fis-4216 194 9 a a DET fis-4216 194 10 successful successful ADJ fis-4216 194 11 first first ADJ fis-4216 194 12 step step NOUN fis-4216 194 13 in in ADP fis-4216 194 14 automatically automatically ADV fis-4216 194 15 detecting detect VERB fis-4216 194 16 concrete concrete ADJ fis-4216 194 17 cracks crack NOUN fis-4216 194 18 from from ADP fis-4216 194 19 captured capture VERB fis-4216 194 20 images image NOUN fis-4216 194 21 . . PUNCT fis-4216 195 1 the the DET fis-4216 195 2 limitations limitation NOUN fis-4216 195 3 of of ADP fis-4216 195 4 the the DET fis-4216 195 5 cracksn cracksn NOUN fis-4216 195 6 system system NOUN fis-4216 195 7 are be AUX fis-4216 195 8 worth worth ADJ fis-4216 195 9 mentioning mention VERB fis-4216 195 10 . . PUNCT fis-4216 196 1 first first ADV fis-4216 196 2 , , PUNCT fis-4216 196 3 our our PRON fis-4216 196 4 model model NOUN fis-4216 196 5 is be AUX fis-4216 196 6 a a DET fis-4216 196 7 patch patch ADJ fis-4216 196 8 - - PUNCT fis-4216 196 9 level level NOUN fis-4216 196 10 crack crack NOUN fis-4216 196 11 detection detection NOUN fis-4216 196 12 system system NOUN fis-4216 196 13 to to PART fis-4216 196 14 identify identify VERB fis-4216 196 15 whether whether SCONJ fis-4216 196 16 there there PRON fis-4216 196 17 is be VERB fis-4216 196 18 a a DET fis-4216 196 19 crack crack NOUN fis-4216 196 20 or or CCONJ fis-4216 196 21 not not PART fis-4216 196 22 within within ADP fis-4216 196 23 a a DET fis-4216 196 24 given give VERB fis-4216 196 25 image image NOUN fis-4216 196 26 . . PUNCT fis-4216 197 1 the the DET fis-4216 197 2 details detail NOUN fis-4216 197 3 of of ADP fis-4216 197 4 the the DET fis-4216 197 5 crack crack NOUN fis-4216 197 6 isolation isolation NOUN fis-4216 197 7 at at ADP fis-4216 197 8 the the DET fis-4216 197 9 pixel pixel PROPN fis-4216 197 10 level level NOUN fis-4216 197 11 are be AUX fis-4216 197 12 obtained obtain VERB fis-4216 197 13 through through ADP fis-4216 197 14 the the DET fis-4216 197 15 following follow VERB fis-4216 197 16 segmentation segmentation NOUN fis-4216 197 17 operation operation NOUN fis-4216 197 18 . . PUNCT fis-4216 198 1 therefore therefore ADV fis-4216 198 2 , , PUNCT fis-4216 198 3 an an DET fis-4216 198 4 implementation implementation NOUN fis-4216 198 5 of of ADP fis-4216 198 6 up up ADV fis-4216 198 7 - - PUNCT fis-4216 198 8 sampling sample VERB fis-4216 198 9 or or CCONJ fis-4216 198 10 deconvolution deconvolution NOUN fis-4216 198 11 layers layer NOUN fis-4216 198 12 would would AUX fis-4216 198 13 be be AUX fis-4216 198 14 embedded embed VERB fis-4216 198 15 later later ADV fis-4216 198 16 in in ADP fis-4216 198 17 the the DET fis-4216 198 18 present present ADJ fis-4216 198 19 model model NOUN fis-4216 198 20 to to PART fis-4216 198 21 establish establish VERB fis-4216 198 22 a a DET fis-4216 198 23 pipeline pipeline NOUN fis-4216 198 24 flow flow NOUN fis-4216 198 25 for for ADP fis-4216 198 26 pixel pixel ADJ fis-4216 198 27 - - PUNCT fis-4216 198 28 level level NOUN fis-4216 198 29 classification classification NOUN fis-4216 198 30 and and CCONJ fis-4216 198 31 crack crack VERB fis-4216 198 32 information information NOUN fis-4216 198 33 determination determination NOUN fis-4216 198 34 [ [ X fis-4216 198 35 3 3 NUM fis-4216 198 36 , , PUNCT fis-4216 198 37 4 4 NUM fis-4216 198 38 ] ] PUNCT fis-4216 198 39 . . PUNCT fis-4216 199 1 secondly secondly ADV fis-4216 199 2 , , PUNCT fis-4216 199 3 the the DET fis-4216 199 4 squeezenet squeezenet NOUN fis-4216 199 5 model model NOUN fis-4216 199 6 can can AUX fis-4216 199 7 only only ADV fis-4216 199 8 operate operate VERB fis-4216 199 9 on on ADP fis-4216 199 10 a a DET fis-4216 199 11 specific specific ADJ fis-4216 199 12 image image NOUN fis-4216 199 13 size size NOUN fis-4216 199 14 of of ADP fis-4216 199 15 227×227 227×227 NUM fis-4216 199 16 , , PUNCT fis-4216 199 17 so so ADV fis-4216 199 18 our our PRON fis-4216 199 19 model model NOUN fis-4216 199 20 is be AUX fis-4216 199 21 highly highly ADV fis-4216 199 22 image image NOUN fis-4216 199 23 size size NOUN fis-4216 199 24 dependent dependent ADJ fis-4216 199 25 . . PUNCT fis-4216 200 1 such such ADJ fis-4216 200 2 dependence dependence NOUN fis-4216 200 3 severely severely ADV fis-4216 200 4 constrains constrain VERB fis-4216 200 5 flexibility flexibility NOUN fis-4216 200 6 by by ADP fis-4216 200 7 taking take VERB fis-4216 200 8 images image NOUN fis-4216 200 9 of of ADP fis-4216 200 10 arbitrary arbitrary ADJ fis-4216 200 11 sizes size NOUN fis-4216 200 12 from from ADP fis-4216 200 13 various various ADJ fis-4216 200 14 imaging imaging NOUN fis-4216 200 15 devices device NOUN fis-4216 200 16 , , PUNCT fis-4216 200 17 such such ADJ fis-4216 200 18 as as ADP fis-4216 200 19 smartphones smartphone NOUN fis-4216 200 20 , , PUNCT fis-4216 200 21 cameras camera NOUN fis-4216 200 22 , , PUNCT fis-4216 200 23 and and CCONJ fis-4216 200 24 uavs uavs NOUN fis-4216 200 25 . . PUNCT fis-4216 201 1 a a DET fis-4216 201 2 fully fully ADV fis-4216 201 3 convolutional convolutional ADJ fis-4216 201 4 network network NOUN fis-4216 201 5 is be AUX fis-4216 201 6 considered consider VERB fis-4216 201 7 an an DET fis-4216 201 8 ideal ideal ADJ fis-4216 201 9 model model NOUN fis-4216 201 10 to to PART fis-4216 201 11 bypass bypass VERB fis-4216 201 12 the the DET fis-4216 201 13 size size NOUN fis-4216 201 14 dependency dependency NOUN fis-4216 201 15 , , PUNCT fis-4216 201 16 enabling enable VERB fis-4216 201 17 the the DET fis-4216 201 18 pixel pixel ADJ fis-4216 201 19 - - PUNCT fis-4216 201 20 level level NOUN fis-4216 201 21 segmentation segmentation NOUN fis-4216 201 22 of of ADP fis-4216 201 23 cracks crack NOUN fis-4216 201 24 [ [ X fis-4216 201 25 4 4 NUM fis-4216 201 26 , , PUNCT fis-4216 201 27 5 5 NUM fis-4216 201 28 , , PUNCT fis-4216 201 29 14 14 NUM fis-4216 201 30 , , PUNCT fis-4216 201 31 22 22 NUM fis-4216 201 32 ] ] PUNCT fis-4216 201 33 . . PUNCT fis-4216 202 1 last last ADJ fis-4216 202 2 but but CCONJ fis-4216 202 3 not not PART fis-4216 202 4 least least ADJ fis-4216 202 5 , , PUNCT fis-4216 202 6 the the DET fis-4216 202 7 training training NOUN fis-4216 202 8 dataset dataset NOUN fis-4216 202 9 does do AUX fis-4216 202 10 not not PART fis-4216 202 11 account account VERB fis-4216 202 12 for for ADP fis-4216 202 13 the the DET fis-4216 202 14 potential potential ADJ fis-4216 202 15 crack crack NOUN fis-4216 202 16 patterns pattern NOUN fis-4216 202 17 , , PUNCT fis-4216 202 18 background background NOUN fis-4216 202 19 material material NOUN fis-4216 202 20 , , PUNCT fis-4216 202 21 texture texture NOUN fis-4216 202 22 , , PUNCT fis-4216 202 23 and and CCONJ fis-4216 202 24 color color NOUN fis-4216 202 25 appearance appearance NOUN fis-4216 202 26 , , PUNCT fis-4216 202 27 which which PRON fis-4216 202 28 can can AUX fis-4216 202 29 cause cause VERB fis-4216 202 30 a a DET fis-4216 202 31 significant significant ADJ fis-4216 202 32 amount amount NOUN fis-4216 202 33 of of ADP fis-4216 202 34 variance variance NOUN fis-4216 202 35 in in ADP fis-4216 202 36 the the DET fis-4216 202 37 captured capture VERB fis-4216 202 38 images image NOUN fis-4216 202 39 of of ADP fis-4216 202 40 the the DET fis-4216 202 41 building building NOUN fis-4216 202 42 surface surface NOUN fis-4216 202 43 that that PRON fis-4216 202 44 are be AUX fis-4216 202 45 recorded record VERB fis-4216 202 46 [ [ X fis-4216 202 47 14 14 NUM fis-4216 202 48 ] ] PUNCT fis-4216 202 49 . . PUNCT fis-4216 203 1 as as ADP fis-4216 203 2 a a DET fis-4216 203 3 result result NOUN fis-4216 203 4 , , PUNCT fis-4216 203 5 it it PRON fis-4216 203 6 will will AUX fis-4216 203 7 be be AUX fis-4216 203 8 necessary necessary ADJ fis-4216 203 9 to to PART fis-4216 203 10 build build VERB fis-4216 203 11 a a DET fis-4216 203 12 versatile versatile ADJ fis-4216 203 13 image image NOUN fis-4216 203 14 dataset dataset NOUN fis-4216 203 15 that that PRON fis-4216 203 16 can can AUX fis-4216 203 17 achieve achieve VERB fis-4216 203 18 realistic realistic ADJ fis-4216 203 19 scenarios scenario NOUN fis-4216 203 20 , , PUNCT fis-4216 203 21 including include VERB fis-4216 203 22 various various ADJ fis-4216 203 23 crack crack NOUN fis-4216 203 24 patterns pattern NOUN fis-4216 203 25 , , PUNCT fis-4216 203 26 crack crack NOUN fis-4216 203 27 - - PUNCT fis-4216 203 28 like like ADJ fis-4216 203 29 distractions distraction NOUN fis-4216 203 30 , , PUNCT fis-4216 203 31 and and CCONJ fis-4216 203 32 background background NOUN fis-4216 203 33 characteristics characteristic NOUN fis-4216 203 34 . . PUNCT fis-4216 204 1 l. l. PROPN fis-4216 204 2 wang wang PROPN fis-4216 204 3 , , PUNCT fis-4216 204 4 frattura frattura PROPN fis-4216 204 5 ed ed PROPN fis-4216 204 6 integrità integrità PROPN fis-4216 204 7 strutturale strutturale PROPN fis-4216 204 8 , , PUNCT fis-4216 204 9 65 65 NUM fis-4216 204 10 ( ( PUNCT fis-4216 204 11 2023 2023 NUM fis-4216 204 12 ) ) PUNCT fis-4216 204 13 289 289 NUM fis-4216 204 14 - - SYM fis-4216 204 15 299 299 NUM fis-4216 204 16 ; ; PUNCT fis-4216 204 17 doi doi NOUN fis-4216 204 18 : : PUNCT fis-4216 204 19 10.3221 10.3221 NUM fis-4216 204 20 / / SYM fis-4216 204 21 igf igf PROPN fis-4216 204 22 - - PUNCT fis-4216 204 23 esis.65.19 esis.65.19 PROPN fis-4216 204 24 297 297 NUM fis-4216 204 25 tn tn NOUN fis-4216 204 26 : : PUNCT fis-4216 204 27 p=0.961 p=0.961 PROPN fis-4216 204 28 tn tn NOUN fis-4216 204 29 : : PUNCT fis-4216 204 30 p=0.907 p=0.907 PROPN fis-4216 204 31 tn tn PROPN fis-4216 204 32 : : PUNCT fis-4216 204 33 p=0.984 p=0.984 PROPN fis-4216 204 34 tn tn PROPN fis-4216 204 35 : : PUNCT fis-4216 204 36 p=0.997 p=0.997 NOUN fis-4216 204 37 tp tp PART fis-4216 204 38 : : PUNCT fis-4216 204 39 p=0.943 p=0.943 PROPN fis-4216 204 40 tp tp NOUN fis-4216 204 41 : : PUNCT fis-4216 204 42 p=0.952 p=0.952 NOUN fis-4216 204 43 i i PRON fis-4216 204 44 m m VERB fis-4216 204 45 ag ag PROPN fis-4216 204 46 e e PROPN fis-4216 204 47 pr pr NOUN fis-4216 204 48 ob ob ADP fis-4216 204 49 ab ab PROPN fis-4216 204 50 ili ili PROPN fis-4216 205 1 ty ty INTJ fis-4216 205 2 tp tp PROPN fis-4216 205 3 : : PUNCT fis-4216 205 4 p=0.993 p=0.993 X fis-4216 205 5 tp tp NOUN fis-4216 205 6 : : PUNCT fis-4216 205 7 p=0.981 p=0.981 NOUN fis-4216 205 8 tp tp ADP fis-4216 205 9 : : PUNCT fis-4216 205 10 p=0.950 p=0.950 PROPN fis-4216 206 1 tp tp PROPN fis-4216 206 2 : : PUNCT fis-4216 206 3 p=0.933 p=0.933 X fis-4216 206 4 tp tp NOUN fis-4216 206 5 : : PUNCT fis-4216 206 6 p=0.969 p=0.969 X fis-4216 206 7 tp tp ADP fis-4216 206 8 : : PUNCT fis-4216 206 9 p=0.994 p=0.994 PUNCT fis-4216 206 10 i i PRON fis-4216 206 11 m m VERB fis-4216 206 12 ag ag PROPN fis-4216 206 13 e e PROPN fis-4216 206 14 pr pr NOUN fis-4216 206 15 ob ob ADP fis-4216 206 16 ab ab PROPN fis-4216 206 17 ili ili PROPN fis-4216 207 1 ty ty INTJ fis-4216 207 2 figure figure NOUN fis-4216 207 3 5 5 NUM fis-4216 207 4 : : PUNCT fis-4216 207 5 detection detection NOUN fis-4216 207 6 of of ADP fis-4216 207 7 concrete concrete ADJ fis-4216 207 8 images image NOUN fis-4216 207 9 and and CCONJ fis-4216 207 10 corresponding correspond VERB fis-4216 207 11 probability probability NOUN fis-4216 207 12 maps map NOUN fis-4216 207 13 . . PUNCT fis-4216 208 1 figure figure VERB fis-4216 208 2 6 6 NUM fis-4216 208 3 : : PUNCT fis-4216 208 4 segmentation segmentation NOUN fis-4216 208 5 of of ADP fis-4216 208 6 concrete concrete ADJ fis-4216 208 7 crack crack NOUN fis-4216 208 8 . . PUNCT fis-4216 209 1 l. l. PROPN fis-4216 209 2 wang wang PROPN fis-4216 209 3 , , PUNCT fis-4216 209 4 frattura frattura PROPN fis-4216 209 5 ed ed PROPN fis-4216 209 6 integrità integrità PROPN fis-4216 209 7 strutturale strutturale PROPN fis-4216 209 8 , , PUNCT fis-4216 209 9 65 65 NUM fis-4216 209 10 ( ( PUNCT fis-4216 209 11 2023 2023 NUM fis-4216 209 12 ) ) PUNCT fis-4216 209 13 289 289 NUM fis-4216 209 14 - - SYM fis-4216 209 15 299 299 NUM fis-4216 209 16 ; ; PUNCT fis-4216 209 17 doi doi NOUN fis-4216 209 18 : : PUNCT fis-4216 209 19 10.3221 10.3221 NUM fis-4216 209 20 / / SYM fis-4216 209 21 igf igf PROPN fis-4216 209 22 - - PUNCT fis-4216 209 23 esis.65.19 esis.65.19 NOUN fis-4216 209 24 298 298 NUM fis-4216 209 25 conclusion conclusion NOUN fis-4216 209 26 racks rack NOUN fis-4216 209 27 in in ADP fis-4216 209 28 concrete concrete ADJ fis-4216 209 29 infrastructure infrastructure NOUN fis-4216 209 30 are be AUX fis-4216 209 31 crucial crucial ADJ fis-4216 209 32 indicators indicator NOUN fis-4216 209 33 of of ADP fis-4216 209 34 structural structural ADJ fis-4216 209 35 damage damage NOUN fis-4216 209 36 . . PUNCT fis-4216 210 1 they they PRON fis-4216 210 2 severely severely ADV fis-4216 210 3 affect affect VERB fis-4216 210 4 structural structural ADJ fis-4216 210 5 integrity integrity NOUN fis-4216 210 6 and and CCONJ fis-4216 210 7 functionality functionality NOUN fis-4216 210 8 . . PUNCT fis-4216 211 1 in in ADP fis-4216 211 2 this this DET fis-4216 211 3 research research NOUN fis-4216 211 4 , , PUNCT fis-4216 211 5 we we PRON fis-4216 211 6 proposed propose VERB fis-4216 211 7 an an DET fis-4216 211 8 automatic automatic ADJ fis-4216 211 9 decision decision NOUN fis-4216 211 10 - - PUNCT fis-4216 211 11 making make VERB fis-4216 211 12 system system NOUN fis-4216 211 13 cracksn cracksn NOUN fis-4216 211 14 based base VERB fis-4216 211 15 on on ADP fis-4216 211 16 deep deep ADJ fis-4216 211 17 learning learning NOUN fis-4216 211 18 to to PART fis-4216 211 19 inspect inspect VERB fis-4216 211 20 cracks crack NOUN fis-4216 211 21 in in ADP fis-4216 211 22 captured capture VERB fis-4216 211 23 digital digital ADJ fis-4216 211 24 images image NOUN fis-4216 211 25 . . PUNCT fis-4216 212 1 the the DET fis-4216 212 2 experimental experimental ADJ fis-4216 212 3 results result NOUN fis-4216 212 4 of of ADP fis-4216 212 5 our our PRON fis-4216 212 6 model model NOUN fis-4216 212 7 provide provide VERB fis-4216 212 8 the the DET fis-4216 212 9 following following ADJ fis-4216 212 10 conclusions conclusion NOUN fis-4216 212 11 : : PUNCT fis-4216 212 12 1 1 X fis-4216 212 13 . . X fis-4216 212 14 through through ADP fis-4216 212 15 the the DET fis-4216 212 16 training training NOUN fis-4216 212 17 and and CCONJ fis-4216 212 18 validation validation NOUN fis-4216 212 19 process process NOUN fis-4216 212 20 , , PUNCT fis-4216 212 21 the the DET fis-4216 212 22 cracksn cracksn NOUN fis-4216 212 23 system system NOUN fis-4216 212 24 automatically automatically ADV fis-4216 212 25 learns learn VERB fis-4216 212 26 the the DET fis-4216 212 27 features feature NOUN fis-4216 212 28 from from ADP fis-4216 212 29 manually manually ADV fis-4216 212 30 annotated annotate VERB fis-4216 212 31 image image NOUN fis-4216 212 32 dataset dataset NOUN fis-4216 212 33 and and CCONJ fis-4216 212 34 successfully successfully ADV fis-4216 212 35 predicts predict VERB fis-4216 212 36 97.3 97.3 NUM fis-4216 212 37 % % NOUN fis-4216 212 38 of of ADP fis-4216 212 39 the the DET fis-4216 212 40 cracked crack VERB fis-4216 212 41 patches patch NOUN fis-4216 212 42 in in ADP fis-4216 212 43 the the DET fis-4216 212 44 test test NOUN fis-4216 212 45 datasets dataset NOUN fis-4216 212 46 . . PUNCT fis-4216 213 1 the the DET fis-4216 213 2 presented present VERB fis-4216 213 3 adam adam NOUN fis-4216 213 4 - - PUNCT fis-4216 213 5 squeezenet squeezenet NOUN fis-4216 213 6 deep deep ADJ fis-4216 213 7 learning learning NOUN fis-4216 213 8 based base VERB fis-4216 213 9 cracksn cracksn NOUN fis-4216 213 10 system system NOUN fis-4216 213 11 exhibits exhibit VERB fis-4216 213 12 superior superior ADJ fis-4216 213 13 performance performance NOUN fis-4216 213 14 in in ADP fis-4216 213 15 crack crack NOUN fis-4216 213 16 detection detection NOUN fis-4216 213 17 compared compare VERB fis-4216 213 18 with with ADP fis-4216 213 19 state state NOUN fis-4216 213 20 - - PUNCT fis-4216 213 21 of of ADP fis-4216 213 22 - - PUNCT fis-4216 213 23 the the DET fis-4216 213 24 - - PUNCT fis-4216 213 25 art art NOUN fis-4216 213 26 models model NOUN fis-4216 213 27 . . PUNCT fis-4216 214 1 2 2 X fis-4216 214 2 . . X fis-4216 214 3 the the DET fis-4216 214 4 deep deep ADJ fis-4216 214 5 adam adam PROPN fis-4216 214 6 - - PUNCT fis-4216 214 7 squeezenet squeezenet NOUN fis-4216 214 8 model model NOUN fis-4216 214 9 with with ADP fis-4216 214 10 a a DET fis-4216 214 11 small small ADJ fis-4216 214 12 structural structural ADJ fis-4216 214 13 size size NOUN fis-4216 214 14 and and CCONJ fis-4216 214 15 superior superior ADJ fis-4216 214 16 performance performance NOUN fis-4216 214 17 is be AUX fis-4216 214 18 preferred prefer VERB fis-4216 214 19 in in ADP fis-4216 214 20 embedded embed VERB fis-4216 214 21 applications application NOUN fis-4216 214 22 for for ADP fis-4216 214 23 real real ADJ fis-4216 214 24 - - PUNCT fis-4216 214 25 time time NOUN fis-4216 214 26 crack crack NOUN fis-4216 214 27 or or CCONJ fis-4216 214 28 damage damage NOUN fis-4216 214 29 detection detection NOUN fis-4216 214 30 in in ADP fis-4216 214 31 concrete concrete ADJ fis-4216 214 32 infrastructure infrastructure NOUN fis-4216 214 33 for for ADP fis-4216 214 34 fast fast ADJ fis-4216 214 35 damage damage NOUN fis-4216 214 36 detection detection NOUN fis-4216 214 37 and and CCONJ fis-4216 214 38 decisionmaking decisionmake VERB fis-4216 214 39 of of ADP fis-4216 214 40 maintenance maintenance NOUN fis-4216 214 41 . . PUNCT fis-4216 215 1 3 3 X fis-4216 215 2 . . X fis-4216 215 3 the the DET fis-4216 215 4 performance performance NOUN fis-4216 215 5 of of ADP fis-4216 215 6 our our PRON fis-4216 215 7 model model NOUN fis-4216 215 8 can can AUX fis-4216 215 9 be be AUX fis-4216 215 10 further far ADV fis-4216 215 11 improved improve VERB fis-4216 215 12 to to ADP fis-4216 215 13 pixel pixel ADJ fis-4216 215 14 - - PUNCT fis-4216 215 15 level level NOUN fis-4216 215 16 accuracy accuracy NOUN fis-4216 215 17 by by ADP fis-4216 215 18 incorporating incorporate VERB fis-4216 215 19 up up ADV fis-4216 215 20 - - PUNCT fis-4216 215 21 sampling sample VERB fis-4216 215 22 layers layer NOUN fis-4216 215 23 or or CCONJ fis-4216 215 24 other other ADJ fis-4216 215 25 deep deep ADJ fis-4216 215 26 learning learning NOUN fis-4216 215 27 algorithm algorithm NOUN fis-4216 215 28 . . PUNCT fis-4216 216 1 training train VERB fis-4216 216 2 the the DET fis-4216 216 3 presented present VERB fis-4216 216 4 model model NOUN fis-4216 216 5 on on ADP fis-4216 216 6 a a DET fis-4216 216 7 wider wide ADJ fis-4216 216 8 variety variety NOUN fis-4216 216 9 of of ADP fis-4216 216 10 image image NOUN fis-4216 216 11 datasets dataset NOUN fis-4216 216 12 with with ADP fis-4216 216 13 realistic realistic ADJ fis-4216 216 14 scenarios scenario NOUN fis-4216 216 15 is be AUX fis-4216 216 16 also also ADV fis-4216 216 17 necessary necessary ADJ fis-4216 216 18 for for ADP fis-4216 216 19 practical practical ADJ fis-4216 216 20 engineering engineering NOUN fis-4216 216 21 applications application NOUN fis-4216 216 22 . . PUNCT fis-4216 217 1 acknowledgment acknowledgment NOUN fis-4216 217 2 he he PRON fis-4216 217 3 financial financial ADJ fis-4216 217 4 support support NOUN fis-4216 217 5 from from ADP fis-4216 217 6 the the DET fis-4216 217 7 university university NOUN fis-4216 217 8 nature nature NOUN fis-4216 217 9 sciences sciences PROPN fis-4216 217 10 research research PROPN fis-4216 217 11 program program NOUN fis-4216 217 12 of of ADP fis-4216 217 13 anhui anhui PROPN fis-4216 217 14 province province PROPN fis-4216 217 15 ( ( PUNCT fis-4216 217 16 kj2021a1001 kj2021a1001 PROPN fis-4216 217 17 ) ) PUNCT fis-4216 217 18 and and CCONJ fis-4216 217 19 the the DET fis-4216 217 20 academic academic ADJ fis-4216 217 21 visit visit NOUN fis-4216 217 22 program program NOUN fis-4216 217 23 of of ADP fis-4216 217 24 anhui anhui PROPN fis-4216 217 25 province province PROPN fis-4216 217 26 ( ( PUNCT fis-4216 217 27 gxgnfx2022054 gxgnfx2022054 PROPN fis-4216 217 28 ) ) PUNCT fis-4216 217 29 is be AUX fis-4216 217 30 greatly greatly ADV fis-4216 217 31 appreciated appreciated ADJ fis-4216 217 32 . . PUNCT fis-4216 218 1 nomenclature nomenclature ADJ fis-4216 218 2 symbol symbol NOUN fis-4216 218 3 meaning mean VERB fis-4216 218 4 symbol symbol NOUN fis-4216 218 5 meaning mean VERB fis-4216 218 6 x x X fis-4216 218 7 matrix matrix NOUN fis-4216 218 8 of of ADP fis-4216 218 9 input input NOUN fis-4216 218 10 image image NOUN fis-4216 218 11 y y PROPN fis-4216 218 12 feature feature NOUN fis-4216 218 13 map map NOUN fis-4216 218 14 w w NOUN fis-4216 218 15 weight weight NOUN fis-4216 218 16 of of ADP fis-4216 218 17 convolution convolution NOUN fis-4216 218 18 filter filter NOUN fis-4216 218 19 z z PROPN fis-4216 218 20 max max PROPN fis-4216 218 21 pool pool NOUN fis-4216 218 22 s s PART fis-4216 218 23 step step NOUN fis-4216 218 24 size size NOUN fis-4216 218 25 of of ADP fis-4216 218 26 stride stride NOUN fis-4216 218 27 f f NOUN fis-4216 218 28 padding padding NOUN fis-4216 218 29 length length NOUN fis-4216 218 30 s s PART fis-4216 218 31 image image NOUN fis-4216 218 32 dataset dataset VERB fis-4216 218 33 p p NOUN fis-4216 218 34 image image NOUN fis-4216 218 35 patch patch NOUN fis-4216 218 36 l l NOUN fis-4216 218 37 class class NOUN fis-4216 218 38 label label NOUN fis-4216 218 39 of of ADP fis-4216 218 40 image image NOUN fis-4216 218 41 patch patch NOUN fis-4216 218 42 q q PROPN fis-4216 218 43 output output NOUN fis-4216 218 44 of of ADP fis-4216 218 45 softmax softmax NOUN fis-4216 218 46 m m PROPN fis-4216 218 47 parameter parameter PROPN fis-4216 218 48 gradient gradient NOUN fis-4216 218 49 v v ADP fis-4216 218 50 squared square VERB fis-4216 218 51 parameter parameter PROPN fis-4216 218 52 gradient gradient PROPN fis-4216 218 53 β1 β1 PROPN fis-4216 218 54 , , PUNCT fis-4216 218 55 β2 β2 NOUN fis-4216 218 56 decay decay VERB fis-4216 218 57 rates rate NOUN fis-4216 218 58 l l NOUN fis-4216 218 59 iteration iteration NOUN fis-4216 218 60 number number NOUN fis-4216 218 61 θ θ PROPN fis-4216 218 62 parameter parameter NOUN fis-4216 218 63 vector vector NOUN fis-4216 218 64 e(θ e(θ ADJ fis-4216 218 65 ) ) PUNCT fis-4216 218 66 loss loss NOUN fis-4216 218 67 function function VERB fis-4216 218 68 γ γ X fis-4216 218 69 rotational rotational ADJ fis-4216 218 70 angle angle NOUN fis-4216 218 71 of of ADP fis-4216 218 72 image image NOUN fis-4216 218 73 patch patch NOUN fis-4216 218 74 α α PROPN fis-4216 218 75 learning learning NOUN fis-4216 218 76 rate rate NOUN fis-4216 218 77 reference reference NOUN fis-4216 218 78 [ [ X fis-4216 218 79 1 1 NUM fis-4216 218 80 ] ] X fis-4216 218 81 sony sony PROPN fis-4216 218 82 , , PUNCT fis-4216 218 83 s. s. PROPN fis-4216 218 84 , , PUNCT fis-4216 218 85 laventure laventure NOUN fis-4216 218 86 , , PUNCT fis-4216 218 87 s. s. PROPN fis-4216 218 88 , , PUNCT fis-4216 218 89 sadhu sadhu PROPN fis-4216 218 90 , , PUNCT fis-4216 218 91 a. a. NOUN fis-4216 218 92 , , PUNCT fis-4216 218 93 ( ( PUNCT fis-4216 218 94 2019 2019 NUM fis-4216 218 95 ) ) PUNCT fis-4216 218 96 . . PUNCT fis-4216 219 1 a a DET fis-4216 219 2 literature literature NOUN fis-4216 219 3 review review NOUN fis-4216 219 4 of of ADP fis-4216 219 5 next next ADJ fis-4216 219 6 - - PUNCT fis-4216 219 7 generation generation NOUN fis-4216 219 8 smart smart ADJ fis-4216 219 9 sensing sense VERB fis-4216 219 10 technology technology NOUN fis-4216 219 11 in in ADP fis-4216 219 12 structural structural ADJ fis-4216 219 13 health health NOUN fis-4216 219 14 monitoring monitoring NOUN fis-4216 219 15 . . PUNCT fis-4216 220 1 struct struct NOUN fis-4216 220 2 . . PUNCT fis-4216 221 1 control control PROPN fis-4216 221 2 health health PROPN fis-4216 221 3 monit monit PROPN fis-4216 221 4 . . PUNCT fis-4216 221 5 , , PUNCT fis-4216 221 6 26(3 26(3 NUM fis-4216 221 7 ) ) PUNCT fis-4216 221 8 , , PUNCT fis-4216 221 9 e2321 e2321 PROPN fis-4216 221 10 . . PUNCT fis-4216 221 11 doi doi NOUN fis-4216 221 12 : : PUNCT fis-4216 221 13 10.1002 10.1002 NUM fis-4216 221 14 / / SYM fis-4216 221 15 stc.2321 stc.2321 PRON fis-4216 221 16 [ [ X fis-4216 221 17 2 2 NUM fis-4216 221 18 ] ] PUNCT fis-4216 221 19 ye ye NUM fis-4216 221 20 , , PUNCT fis-4216 221 21 x. x. PROPN fis-4216 221 22 w. w. PROPN fis-4216 221 23 , , PUNCT fis-4216 221 24 jin jin PROPN fis-4216 221 25 , , PUNCT fis-4216 221 26 t. t. PROPN fis-4216 221 27 , , PUNCT fis-4216 221 28 yun yun PROPN fis-4216 221 29 , , PUNCT fis-4216 221 30 c. c. PROPN fis-4216 221 31 b. b. PROPN fis-4216 221 32 , , PUNCT fis-4216 221 33 ( ( PUNCT fis-4216 221 34 2019 2019 NUM fis-4216 221 35 ) ) PUNCT fis-4216 221 36 . . PUNCT fis-4216 222 1 a a DET fis-4216 222 2 review review NOUN fis-4216 222 3 on on ADP fis-4216 222 4 deep deep ADJ fis-4216 222 5 learning learning NOUN fis-4216 222 6 - - PUNCT fis-4216 222 7 based base VERB fis-4216 222 8 structural structural ADJ fis-4216 222 9 health health NOUN fis-4216 222 10 monitoring monitoring NOUN fis-4216 222 11 of of ADP fis-4216 222 12 civil civil ADJ fis-4216 222 13 infrastructures infrastructure NOUN fis-4216 222 14 . . PUNCT fis-4216 223 1 smart smart ADJ fis-4216 223 2 struct struct NOUN fis-4216 223 3 . . PUNCT fis-4216 224 1 syst syst PROPN fis-4216 224 2 . . PROPN fis-4216 224 3 , , PUNCT fis-4216 224 4 24(5 24(5 NOUN fis-4216 224 5 ) ) PUNCT fis-4216 224 6 , , PUNCT fis-4216 224 7 pp pp PROPN fis-4216 224 8 . . PUNCT fis-4216 225 1 567 567 NUM fis-4216 225 2 - - SYM fis-4216 225 3 585 585 NUM fis-4216 225 4 . . PUNCT fis-4216 226 1 doi doi NOUN fis-4216 226 2 : : PUNCT fis-4216 226 3 10.12989 10.12989 NUM fis-4216 226 4 / / SYM fis-4216 226 5 sss.2019.24.5.567 sss.2019.24.5.567 NOUN fis-4216 226 6 . . PUNCT fis-4216 227 1 [ [ X fis-4216 227 2 3 3 NUM fis-4216 227 3 ] ] X fis-4216 227 4 fei fei PROPN fis-4216 227 5 , , PUNCT fis-4216 227 6 y. y. PROPN fis-4216 227 7 , , PUNCT fis-4216 227 8 wang wang PROPN fis-4216 227 9 , , PUNCT fis-4216 227 10 k. k. PROPN fis-4216 227 11 c. c. PROPN fis-4216 227 12 , , PUNCT fis-4216 227 13 zhang zhang PROPN fis-4216 227 14 , , PUNCT fis-4216 227 15 a. a. PROPN fis-4216 227 16 , , PUNCT fis-4216 227 17 chen chen PROPN fis-4216 227 18 , , PUNCT fis-4216 227 19 c. c. PROPN fis-4216 227 20 , , PUNCT fis-4216 227 21 li li PROPN fis-4216 227 22 , , PUNCT fis-4216 227 23 j. j. PROPN fis-4216 227 24 q. q. PROPN fis-4216 227 25 , , PUNCT fis-4216 227 26 liu liu PROPN fis-4216 227 27 , , PUNCT fis-4216 227 28 y. y. PROPN fis-4216 227 29 , , PUNCT fis-4216 227 30 li li PROPN fis-4216 227 31 , , PUNCT fis-4216 227 32 b. b. PROPN fis-4216 227 33 , , PUNCT fis-4216 227 34 ( ( PUNCT fis-4216 227 35 2019 2019 NUM fis-4216 227 36 ) ) PUNCT fis-4216 227 37 . . PUNCT fis-4216 228 1 pixel pixel ADJ fis-4216 228 2 - - PUNCT fis-4216 228 3 level level NOUN fis-4216 228 4 cracking crack VERB fis-4216 228 5 detection detection NOUN fis-4216 228 6 on on ADP fis-4216 228 7 3d 3d NUM fis-4216 228 8 asphalt asphalt NOUN fis-4216 228 9 pavement pavement NOUN fis-4216 228 10 images image NOUN fis-4216 228 11 through through ADP fis-4216 228 12 deep deep ADJ fis-4216 228 13 - - PUNCT fis-4216 228 14 learning learning NOUN fis-4216 228 15 - - PUNCT fis-4216 228 16 based base VERB fis-4216 228 17 cracknet cracknet NOUN fis-4216 228 18 - - PUNCT fis-4216 228 19 v. v. PROPN fis-4216 228 20 ieee ieee PROPN fis-4216 228 21 trans trans PROPN fis-4216 228 22 . . PUNCT fis-4216 229 1 intell intell PROPN fis-4216 229 2 . . PUNCT fis-4216 230 1 transp transp PROPN fis-4216 230 2 . . PUNCT fis-4216 231 1 syst syst PROPN fis-4216 231 2 . . PROPN fis-4216 231 3 , , PUNCT fis-4216 231 4 21(1 21(1 NUM fis-4216 231 5 ) ) PUNCT fis-4216 231 6 , , PUNCT fis-4216 231 7 pp pp PROPN fis-4216 231 8 . . PUNCT fis-4216 232 1 273 273 NUM fis-4216 232 2 - - SYM fis-4216 232 3 284 284 NUM fis-4216 232 4 . . PUNCT fis-4216 233 1 doi doi NOUN fis-4216 233 2 : : PUNCT fis-4216 233 3 10.1109 10.1109 NUM fis-4216 233 4 / / SYM fis-4216 233 5 tits.2019.2891167 tits.2019.2891167 NUM fis-4216 233 6 . . PUNCT fis-4216 234 1 [ [ X fis-4216 234 2 4 4 NUM fis-4216 234 3 ] ] X fis-4216 234 4 alipour alipour ADJ fis-4216 234 5 , , PUNCT fis-4216 234 6 m. m. NOUN fis-4216 234 7 , , PUNCT fis-4216 234 8 harris harris PROPN fis-4216 234 9 , , PUNCT fis-4216 234 10 d. d. PROPN fis-4216 234 11 k. k. PROPN fis-4216 234 12 , , PUNCT fis-4216 234 13 and and CCONJ fis-4216 234 14 miller miller PROPN fis-4216 234 15 , , PUNCT fis-4216 234 16 g. g. PROPN fis-4216 234 17 r. r. PROPN fis-4216 234 18 , , PUNCT fis-4216 234 19 ( ( PUNCT fis-4216 234 20 2019 2019 NUM fis-4216 234 21 ) ) PUNCT fis-4216 234 22 . . PUNCT fis-4216 235 1 robust robust ADJ fis-4216 235 2 pixel pixel ADJ fis-4216 235 3 - - PUNCT fis-4216 235 4 level level NOUN fis-4216 235 5 crack crack NOUN fis-4216 235 6 detection detection NOUN fis-4216 235 7 using use VERB fis-4216 235 8 deep deep ADV fis-4216 235 9 fully fully ADV fis-4216 235 10 convolutional convolutional ADJ fis-4216 235 11 neural neural ADJ fis-4216 235 12 networks network NOUN fis-4216 235 13 . . PUNCT fis-4216 236 1 j. j. PROPN fis-4216 236 2 comput comput PROPN fis-4216 236 3 . . PUNCT fis-4216 237 1 civ civ NOUN fis-4216 237 2 . . PUNCT fis-4216 238 1 eng eng PROPN fis-4216 238 2 . . PROPN fis-4216 238 3 , , PUNCT fis-4216 238 4 33(6 33(6 PROPN fis-4216 238 5 ) ) PUNCT fis-4216 238 6 , , PUNCT fis-4216 238 7 04019040 04019040 NUM fis-4216 238 8 . . PUNCT fis-4216 239 1 doi doi NOUN fis-4216 239 2 : : PUNCT fis-4216 239 3 10.1061/(asce)cp.1943 10.1061/(asce)cp.1943 NUM fis-4216 239 4 - - PUNCT fis-4216 239 5 5487.0000854 5487.0000854 NUM fis-4216 239 6 . . PUNCT fis-4216 240 1 [ [ X fis-4216 240 2 5 5 NUM fis-4216 240 3 ] ] SYM fis-4216 240 4 ni ni PROPN fis-4216 240 5 , , PUNCT fis-4216 240 6 f. f. PROPN fis-4216 240 7 , , PUNCT fis-4216 240 8 zhang zhang PROPN fis-4216 240 9 , , PUNCT fis-4216 240 10 j. j. PROPN fis-4216 240 11 , , PUNCT fis-4216 240 12 chen chen PROPN fis-4216 240 13 , , PUNCT fis-4216 240 14 z. z. PROPN fis-4216 240 15 , , PUNCT fis-4216 240 16 ( ( PUNCT fis-4216 240 17 2019 2019 NUM fis-4216 240 18 ) ) PUNCT fis-4216 240 19 . . PUNCT fis-4216 241 1 pixel‐level pixel‐level NOUN fis-4216 241 2 crack crack VERB fis-4216 241 3 delineation delineation NOUN fis-4216 241 4 in in ADP fis-4216 241 5 images image NOUN fis-4216 241 6 with with ADP fis-4216 241 7 convolutional convolutional ADJ fis-4216 241 8 feature feature NOUN fis-4216 241 9 fusion fusion NOUN fis-4216 241 10 . . PUNCT fis-4216 242 1 struct struct NOUN fis-4216 242 2 . . PUNCT fis-4216 243 1 control control PROPN fis-4216 243 2 health health PROPN fis-4216 243 3 monit monit PROPN fis-4216 243 4 . . PUNCT fis-4216 243 5 , , PUNCT fis-4216 243 6 26(1 26(1 NUM fis-4216 243 7 ) ) PUNCT fis-4216 244 1 , , PUNCT fis-4216 244 2 e2286 e2286 PROPN fis-4216 244 3 . . PUNCT fis-4216 244 4 doi doi NOUN fis-4216 244 5 : : PUNCT fis-4216 244 6 10.1002 10.1002 NUM fis-4216 244 7 / / SYM fis-4216 244 8 stc.2286 stc.2286 NOUN fis-4216 244 9 . . PUNCT fis-4216 245 1 c c PROPN fis-4216 245 2 t t PROPN fis-4216 245 3 l. l. PROPN fis-4216 245 4 wang wang PROPN fis-4216 245 5 , , PUNCT fis-4216 245 6 frattura frattura PROPN fis-4216 245 7 ed ed PROPN fis-4216 245 8 integrità integrità PROPN fis-4216 245 9 strutturale strutturale PROPN fis-4216 245 10 , , PUNCT fis-4216 245 11 65 65 NUM fis-4216 245 12 ( ( PUNCT fis-4216 245 13 2023 2023 NUM fis-4216 245 14 ) ) PUNCT fis-4216 245 15 289 289 NUM fis-4216 245 16 - - SYM fis-4216 245 17 299 299 NUM fis-4216 245 18 ; ; PUNCT fis-4216 245 19 doi doi NOUN fis-4216 245 20 : : PUNCT fis-4216 245 21 10.3221 10.3221 NUM fis-4216 245 22 / / SYM fis-4216 245 23 igf igf PROPN fis-4216 245 24 - - PUNCT fis-4216 245 25 esis.65.19 esis.65.19 NOUN fis-4216 245 26 299 299 NUM fis-4216 245 27 [ [ X fis-4216 245 28 6 6 NUM fis-4216 245 29 ] ] X fis-4216 245 30 zhao zhao PROPN fis-4216 245 31 , , PUNCT fis-4216 245 32 s. s. PROPN fis-4216 245 33 , , PUNCT fis-4216 245 34 kang kang PROPN fis-4216 245 35 , , PUNCT fis-4216 245 36 f. f. PROPN fis-4216 245 37 , , PUNCT fis-4216 245 38 li li PROPN fis-4216 245 39 , , PUNCT fis-4216 245 40 j. j. PROPN fis-4216 245 41 , , PUNCT fis-4216 245 42 ma ma PROPN fis-4216 245 43 , , PUNCT fis-4216 245 44 c. c. PROPN fis-4216 245 45 , , PUNCT fis-4216 245 46 ( ( PUNCT fis-4216 245 47 2021 2021 NUM fis-4216 245 48 ) ) PUNCT fis-4216 245 49 . . PUNCT fis-4216 246 1 structural structural ADJ fis-4216 246 2 health health NOUN fis-4216 246 3 monitoring monitoring NOUN fis-4216 246 4 and and CCONJ fis-4216 246 5 inspection inspection NOUN fis-4216 246 6 of of ADP fis-4216 246 7 dams dam NOUN fis-4216 246 8 based base VERB fis-4216 246 9 on on ADP fis-4216 246 10 uav uav PROPN fis-4216 246 11 photogrammetry photogrammetry NOUN fis-4216 246 12 with with ADP fis-4216 246 13 image image NOUN fis-4216 246 14 3d 3d PROPN fis-4216 246 15 reconstruction reconstruction NOUN fis-4216 246 16 . . PUNCT fis-4216 247 1 autom autom PROPN fis-4216 247 2 . . PUNCT fis-4216 248 1 constr constr PROPN fis-4216 248 2 . . PROPN fis-4216 248 3 , , PUNCT fis-4216 248 4 130 130 NUM fis-4216 248 5 , , PUNCT fis-4216 248 6 103832 103832 NUM fis-4216 248 7 . . PUNCT fis-4216 249 1 doi doi NOUN fis-4216 249 2 : : PUNCT fis-4216 249 3 10.1016 10.1016 NUM fis-4216 249 4 / / SYM fis-4216 249 5 j.autcon.2021.103832 j.autcon.2021.103832 NOUN fis-4216 249 6 . . PUNCT fis-4216 250 1 [ [ X fis-4216 250 2 7 7 NUM fis-4216 250 3 ] ] X fis-4216 250 4 yan yan PROPN fis-4216 250 5 , , PUNCT fis-4216 250 6 y. y. PROPN fis-4216 250 7 , , PUNCT fis-4216 250 8 mao mao PROPN fis-4216 250 9 , , PUNCT fis-4216 250 10 z. z. PROPN fis-4216 250 11 , , PUNCT fis-4216 250 12 wu wu PROPN fis-4216 250 13 , , PUNCT fis-4216 250 14 j. j. PROPN fis-4216 250 15 , , PUNCT fis-4216 250 16 padir padir NOUN fis-4216 250 17 , , PUNCT fis-4216 250 18 t. t. PROPN fis-4216 250 19 , , PUNCT fis-4216 250 20 hajjar hajjar PROPN fis-4216 250 21 , , PUNCT fis-4216 250 22 j. j. PROPN fis-4216 250 23 f. f. PROPN fis-4216 250 24 , , PUNCT fis-4216 250 25 ( ( PUNCT fis-4216 250 26 2021 2021 NUM fis-4216 250 27 ) ) PUNCT fis-4216 250 28 . . PUNCT fis-4216 251 1 towards towards ADP fis-4216 251 2 automated automate VERB fis-4216 251 3 detection detection NOUN fis-4216 251 4 and and CCONJ fis-4216 251 5 quantification quantification NOUN fis-4216 251 6 of of ADP fis-4216 251 7 concrete concrete ADJ fis-4216 251 8 cracks crack NOUN fis-4216 251 9 using use VERB fis-4216 251 10 integrated integrate VERB fis-4216 251 11 images image NOUN fis-4216 251 12 and and CCONJ fis-4216 251 13 lidar lidar NOUN fis-4216 251 14 data datum NOUN fis-4216 251 15 from from ADP fis-4216 251 16 unmanned unmanned ADJ fis-4216 251 17 aerial aerial ADJ fis-4216 251 18 vehicles vehicle NOUN fis-4216 251 19 . . PUNCT fis-4216 252 1 struct struct NOUN fis-4216 252 2 . . PUNCT fis-4216 253 1 control control PROPN fis-4216 253 2 health health PROPN fis-4216 253 3 monit monit PROPN fis-4216 253 4 . . PUNCT fis-4216 253 5 , , PUNCT fis-4216 253 6 28(8 28(8 NUM fis-4216 253 7 ) ) PUNCT fis-4216 253 8 , , PUNCT fis-4216 253 9 e2757 e2757 PROPN fis-4216 253 10 . . PUNCT fis-4216 253 11 doi doi NOUN fis-4216 253 12 : : PUNCT fis-4216 253 13 10.1002 10.1002 NUM fis-4216 253 14 / / SYM fis-4216 253 15 stc.2757 stc.2757 PROPN fis-4216 253 16 . . PUNCT fis-4216 254 1 [ [ X fis-4216 254 2 8 8 NUM fis-4216 254 3 ] ] X fis-4216 254 4 ellenberg ellenberg PROPN fis-4216 254 5 , , PUNCT fis-4216 254 6 a. a. NOUN fis-4216 254 7 , , PUNCT fis-4216 254 8 kontsos kontsos PROPN fis-4216 254 9 , , PUNCT fis-4216 254 10 a. a. NOUN fis-4216 254 11 , , PUNCT fis-4216 254 12 moon moon PROPN fis-4216 254 13 , , PUNCT fis-4216 254 14 f. f. PROPN fis-4216 254 15 , , PUNCT fis-4216 254 16 & & CCONJ fis-4216 254 17 bartoli bartoli PROPN fis-4216 254 18 , , PUNCT fis-4216 254 19 i. i. PROPN fis-4216 254 20 ( ( PUNCT fis-4216 254 21 2016 2016 NUM fis-4216 254 22 ) ) PUNCT fis-4216 254 23 . . PUNCT fis-4216 255 1 bridge bridge NOUN fis-4216 255 2 related relate VERB fis-4216 255 3 damage damage NOUN fis-4216 255 4 quantification quantification NOUN fis-4216 255 5 using use VERB fis-4216 255 6 unmanned unmanned ADJ fis-4216 255 7 aerial aerial ADJ fis-4216 255 8 vehicle vehicle NOUN fis-4216 255 9 imagery imagery NOUN fis-4216 255 10 . . PUNCT fis-4216 256 1 struct struct NOUN fis-4216 256 2 . . PUNCT fis-4216 257 1 control control PROPN fis-4216 257 2 health health PROPN fis-4216 257 3 monit monit PROPN fis-4216 257 4 . . PROPN fis-4216 257 5 , , PUNCT fis-4216 257 6 23(9 23(9 NUM fis-4216 257 7 ) ) PUNCT fis-4216 257 8 , , PUNCT fis-4216 257 9 pp pp ADJ fis-4216 257 10 . . PUNCT fis-4216 258 1 1168 1168 NUM fis-4216 258 2 - - SYM fis-4216 258 3 1179 1179 NUM fis-4216 258 4 . . PUNCT fis-4216 259 1 doi doi NOUN fis-4216 259 2 : : PUNCT fis-4216 259 3 10.1002 10.1002 NUM fis-4216 259 4 / / SYM fis-4216 259 5 stc.1831 stc.1831 PROPN fis-4216 259 6 . . PUNCT fis-4216 260 1 [ [ X fis-4216 260 2 9 9 NUM fis-4216 260 3 ] ] SYM fis-4216 260 4 dorafshan dorafshan NOUN fis-4216 260 5 , , PUNCT fis-4216 260 6 s. s. PROPN fis-4216 260 7 , , PUNCT fis-4216 260 8 thomas thomas PROPN fis-4216 260 9 , , PUNCT fis-4216 260 10 r. r. PROPN fis-4216 260 11 j. j. PROPN fis-4216 260 12 , , PUNCT fis-4216 260 13 maguire maguire PROPN fis-4216 260 14 , , PUNCT fis-4216 260 15 m. m. NOUN fis-4216 260 16 , , PUNCT fis-4216 260 17 ( ( PUNCT fis-4216 260 18 2018 2018 NUM fis-4216 260 19 ) ) PUNCT fis-4216 260 20 . . PUNCT fis-4216 261 1 comparison comparison NOUN fis-4216 261 2 of of ADP fis-4216 261 3 deep deep ADJ fis-4216 261 4 convolutional convolutional ADJ fis-4216 261 5 neural neural ADJ fis-4216 261 6 networks network NOUN fis-4216 261 7 and and CCONJ fis-4216 261 8 edge edge NOUN fis-4216 261 9 detectors detector NOUN fis-4216 261 10 for for ADP fis-4216 261 11 image image NOUN fis-4216 261 12 - - PUNCT fis-4216 261 13 based base VERB fis-4216 261 14 crack crack NOUN fis-4216 261 15 detection detection NOUN fis-4216 261 16 in in ADP fis-4216 261 17 concrete concrete NOUN fis-4216 261 18 . . PUNCT fis-4216 262 1 constr constr NOUN fis-4216 262 2 . . PUNCT fis-4216 263 1 build build PROPN fis-4216 263 2 mater mater NOUN fis-4216 263 3 . . PROPN fis-4216 263 4 , , PUNCT fis-4216 263 5 186 186 NUM fis-4216 263 6 , , PUNCT fis-4216 263 7 pp pp ADJ fis-4216 263 8 . . PUNCT fis-4216 264 1 1031 1031 NUM fis-4216 264 2 - - PUNCT fis-4216 264 3 1045 1045 NUM fis-4216 264 4 . . PUNCT fis-4216 265 1 doi doi NOUN fis-4216 265 2 : : PUNCT fis-4216 265 3 10.1016/ 10.1016/ NUM fis-4216 265 4 j.conbuildmat.2018.08.011 j.conbuildmat.2018.08.011 NOUN fis-4216 265 5 . . PUNCT fis-4216 266 1 [ [ X fis-4216 266 2 10 10 NUM fis-4216 266 3 ] ] X fis-4216 266 4 gurunsi gurunsi NOUN fis-4216 266 5 , , PUNCT fis-4216 266 6 n. n. PROPN fis-4216 266 7 , , PUNCT fis-4216 266 8 b. b. PROPN fis-4216 266 9 basily basily PROPN fis-4216 266 10 , , PUNCT fis-4216 266 11 j. j. PROPN fis-4216 266 12 kim kim PROPN fis-4216 266 13 , , PUNCT fis-4216 266 14 j. j. PROPN fis-4216 266 15 yi yi PROPN fis-4216 266 16 , , PUNCT fis-4216 266 17 t. t. PROPN fis-4216 266 18 duong duong PROPN fis-4216 266 19 , , PUNCT fis-4216 266 20 k. k. PROPN fis-4216 266 21 dinh dinh PROPN fis-4216 266 22 , , PUNCT fis-4216 266 23 s.-h s.-h NOUN fis-4216 266 24 . . PUNCT fis-4216 267 1 kee kee PROPN fis-4216 267 2 , , PUNCT fis-4216 267 3 and and CCONJ fis-4216 267 4 a. a. NOUN fis-4216 267 5 maher maher PROPN fis-4216 267 6 . . PUNCT fis-4216 267 7 , , PUNCT fis-4216 267 8 ( ( PUNCT fis-4216 267 9 2017 2017 NUM fis-4216 267 10 ) ) PUNCT fis-4216 267 11 . . PUNCT fis-4216 268 1 rabit rabit NOUN fis-4216 268 2 : : PUNCT fis-4216 268 3 implementation implementation NOUN fis-4216 268 4 , , PUNCT fis-4216 268 5 performance performance NOUN fis-4216 268 6 validation validation NOUN fis-4216 268 7 and and CCONJ fis-4216 268 8 integration integration NOUN fis-4216 268 9 with with ADP fis-4216 268 10 other other ADJ fis-4216 268 11 robotic robotic ADJ fis-4216 268 12 platforms platform NOUN fis-4216 268 13 for for ADP fis-4216 268 14 improved improved ADJ fis-4216 268 15 management management NOUN fis-4216 268 16 of of ADP fis-4216 268 17 bridge bridge NOUN fis-4216 268 18 decks deck NOUN fis-4216 268 19 . . PUNCT fis-4216 269 1 int int NOUN fis-4216 269 2 . . PUNCT fis-4216 270 1 j. j. PROPN fis-4216 270 2 intell intell PROPN fis-4216 270 3 . . PUNCT fis-4216 271 1 rob rob PROPN fis-4216 271 2 . . PUNCT fis-4216 272 1 appl appl PROPN fis-4216 272 2 . . PROPN fis-4216 272 3 , , PUNCT fis-4216 272 4 1(3 1(3 NUM fis-4216 272 5 ) ) PUNCT fis-4216 272 6 , , PUNCT fis-4216 272 7 pp pp ADP fis-4216 272 8 . . PUNCT fis-4216 273 1 271–286 271–286 NUM fis-4216 273 2 . . PUNCT fis-4216 274 1 doi doi NOUN fis-4216 274 2 : : PUNCT fis-4216 274 3 10.1007 10.1007 NUM fis-4216 274 4 / / SYM fis-4216 274 5 s41315 s41315 NOUN fis-4216 274 6 - - PUNCT fis-4216 274 7 017 017 NUM fis-4216 274 8 - - PUNCT fis-4216 274 9 0027 0027 NUM fis-4216 274 10 - - SYM fis-4216 274 11 5 5 NUM fis-4216 274 12 . . PUNCT fis-4216 275 1 [ [ X fis-4216 275 2 11 11 NUM fis-4216 275 3 ] ] X fis-4216 275 4 zhang zhang PROPN fis-4216 275 5 , , PUNCT fis-4216 275 6 l. l. PROPN fis-4216 275 7 , , PUNCT fis-4216 275 8 yang yang PROPN fis-4216 275 9 , , PUNCT fis-4216 275 10 f. f. PROPN fis-4216 275 11 , , PUNCT fis-4216 275 12 zhang zhang PROPN fis-4216 275 13 , , PUNCT fis-4216 275 14 y. y. PROPN fis-4216 275 15 d. d. PROPN fis-4216 275 16 , , PUNCT fis-4216 275 17 zhu zhu PROPN fis-4216 275 18 , , PUNCT fis-4216 275 19 y. y. PROPN fis-4216 275 20 j. j. PROPN fis-4216 275 21 , , PUNCT fis-4216 275 22 ( ( PUNCT fis-4216 275 23 2016 2016 NUM fis-4216 275 24 ) ) PUNCT fis-4216 275 25 . . PUNCT fis-4216 276 1 road road NOUN fis-4216 276 2 crack crack NOUN fis-4216 276 3 detection detection NOUN fis-4216 276 4 using use VERB fis-4216 276 5 deep deep ADJ fis-4216 276 6 convolutional convolutional ADJ fis-4216 276 7 neural neural ADJ fis-4216 276 8 network network NOUN fis-4216 276 9 . . PUNCT fis-4216 277 1 in in ADP fis-4216 277 2 2016 2016 NUM fis-4216 277 3 ieee ieee NOUN fis-4216 277 4 international international ADJ fis-4216 277 5 conference conference NOUN fis-4216 277 6 on on ADP fis-4216 277 7 image image NOUN fis-4216 277 8 processing processing NOUN fis-4216 277 9 ( ( PUNCT fis-4216 277 10 icip icip PROPN fis-4216 277 11 ) ) PUNCT fis-4216 277 12 , , PUNCT fis-4216 277 13 pp pp ADP fis-4216 277 14 . . PUNCT fis-4216 278 1 3708 3708 NUM fis-4216 278 2 - - PUNCT fis-4216 278 3 3712 3712 NUM fis-4216 278 4 . . PUNCT fis-4216 279 1 ieee ieee NOUN fis-4216 279 2 . . PUNCT fis-4216 280 1 doi doi PROPN fis-4216 280 2 : : PUNCT fis-4216 280 3 10.1109/ 10.1109/ NUM fis-4216 280 4 icip.2016.7533052 icip.2016.7533052 NOUN fis-4216 280 5 . . PUNCT fis-4216 281 1 [ [ X fis-4216 281 2 12 12 NUM fis-4216 281 3 ] ] X fis-4216 281 4 cha cha X fis-4216 281 5 , , PUNCT fis-4216 281 6 y. y. PROPN fis-4216 281 7 j. j. PROPN fis-4216 281 8 , , PUNCT fis-4216 281 9 w. w. PROPN fis-4216 281 10 choi choi PROPN fis-4216 281 11 , , PUNCT fis-4216 281 12 o. o. PROPN fis-4216 281 13 büyüköztürk büyüköztürk PROPN fis-4216 281 14 . . PROPN fis-4216 281 15 , , PUNCT fis-4216 281 16 ( ( PUNCT fis-4216 281 17 2017 2017 NUM fis-4216 281 18 ) ) PUNCT fis-4216 281 19 . . PUNCT fis-4216 282 1 deep deep ADJ fis-4216 282 2 learning learning NOUN fis-4216 282 3 based base VERB fis-4216 282 4 crack crack NOUN fis-4216 282 5 damage damage NOUN fis-4216 282 6 detection detection NOUN fis-4216 282 7 using use VERB fis-4216 282 8 convolutional convolutional ADJ fis-4216 282 9 neural neural ADJ fis-4216 282 10 networks network NOUN fis-4216 282 11 . . PUNCT fis-4216 283 1 comput.-aided comput.-aided ADJ fis-4216 283 2 civ civ NOUN fis-4216 283 3 . . PUNCT fis-4216 283 4 infrastruct infrastruct NOUN fis-4216 283 5 . . PUNCT fis-4216 284 1 eng eng PROPN fis-4216 284 2 . . PROPN fis-4216 284 3 , , PUNCT fis-4216 284 4 32(5 32(5 NUM fis-4216 284 5 ) ) PUNCT fis-4216 284 6 , , PUNCT fis-4216 284 7 pp pp ADP fis-4216 284 8 . . PUNCT fis-4216 285 1 361–378 361–378 NUM fis-4216 285 2 . . PUNCT fis-4216 285 3 doi doi NOUN fis-4216 285 4 : : PUNCT fis-4216 285 5 10.1111 10.1111 NUM fis-4216 285 6 / / SYM fis-4216 285 7 mice.12263 mice.12263 PROPN fis-4216 285 8 . . PUNCT fis-4216 286 1 [ [ X fis-4216 286 2 13 13 NUM fis-4216 286 3 ] ] X fis-4216 286 4 chen chen PROPN fis-4216 286 5 , , PUNCT fis-4216 286 6 j. j. PROPN fis-4216 286 7 , , PUNCT fis-4216 286 8 he he PRON fis-4216 286 9 , , PUNCT fis-4216 286 10 y. y. PROPN fis-4216 286 11 , , PUNCT fis-4216 286 12 ( ( PUNCT fis-4216 286 13 2022 2022 NUM fis-4216 286 14 ) ) PUNCT fis-4216 286 15 . . PUNCT fis-4216 287 1 a a DET fis-4216 287 2 novel novel NOUN fis-4216 287 3 u‐shaped u‐shaped SYM fis-4216 287 4 encoder encoder NOUN fis-4216 287 5 – – PUNCT fis-4216 287 6 decoder decoder NOUN fis-4216 287 7 network network NOUN fis-4216 287 8 with with ADP fis-4216 287 9 attention attention NOUN fis-4216 287 10 mechanism mechanism NOUN fis-4216 287 11 for for ADP fis-4216 287 12 detection detection NOUN fis-4216 287 13 and and CCONJ fis-4216 287 14 evaluation evaluation NOUN fis-4216 287 15 of of ADP fis-4216 287 16 road road NOUN fis-4216 287 17 cracks crack NOUN fis-4216 287 18 at at ADP fis-4216 287 19 pixel pixel PROPN fis-4216 287 20 level level NOUN fis-4216 287 21 . . PUNCT fis-4216 288 1 comput.-aidedciv comput.-aidedciv PROPN fis-4216 288 2 . . PROPN fis-4216 288 3 infrastruct infrastruct NOUN fis-4216 288 4 . . PUNCT fis-4216 289 1 eng eng PROPN fis-4216 289 2 . . PROPN fis-4216 289 3 , , PUNCT fis-4216 289 4 37(13 37(13 NUM fis-4216 289 5 ) ) PUNCT fis-4216 289 6 , , PUNCT fis-4216 289 7 pp pp PROPN fis-4216 289 8 . . PUNCT fis-4216 290 1 1721 1721 NUM fis-4216 290 2 - - SYM fis-4216 290 3 1736 1736 NUM fis-4216 290 4 . . PUNCT fis-4216 291 1 doi doi NOUN fis-4216 291 2 : : PUNCT fis-4216 291 3 10.1111/ 10.1111/ PROPN fis-4216 291 4 mice.12826 mice.12826 PROPN fis-4216 291 5 . . PUNCT fis-4216 292 1 [ [ X fis-4216 292 2 14 14 NUM fis-4216 292 3 ] ] X fis-4216 292 4 zhu zhu PROPN fis-4216 292 5 , , PUNCT fis-4216 292 6 j. j. PROPN fis-4216 292 7 , , PUNCT fis-4216 292 8 zhong zhong PROPN fis-4216 292 9 , , PUNCT fis-4216 292 10 j. j. PROPN fis-4216 292 11 , , PUNCT fis-4216 292 12 ma ma PROPN fis-4216 292 13 , , PUNCT fis-4216 292 14 t. t. PROPN fis-4216 292 15 , , PUNCT fis-4216 292 16 huang huang PROPN fis-4216 292 17 , , PUNCT fis-4216 292 18 x. x. PROPN fis-4216 292 19 , , PUNCT fis-4216 292 20 zhang zhang PROPN fis-4216 292 21 , , PUNCT fis-4216 292 22 w. w. PROPN fis-4216 292 23 , , PUNCT fis-4216 292 24 zhou zhou PROPN fis-4216 292 25 , , PUNCT fis-4216 292 26 y. y. PROPN fis-4216 292 27 , , PUNCT fis-4216 292 28 ( ( PUNCT fis-4216 292 29 2022 2022 NUM fis-4216 292 30 ) ) PUNCT fis-4216 292 31 . . PUNCT fis-4216 293 1 pavement pavement NOUN fis-4216 293 2 distress distress NOUN fis-4216 293 3 detection detection NOUN fis-4216 293 4 using use VERB fis-4216 293 5 convolutional convolutional ADJ fis-4216 293 6 neural neural ADJ fis-4216 293 7 networks network NOUN fis-4216 293 8 with with ADP fis-4216 293 9 images image NOUN fis-4216 293 10 captured capture VERB fis-4216 293 11 via via ADP fis-4216 293 12 uav uav PROPN fis-4216 293 13 . . PUNCT fis-4216 294 1 autom autom PROPN fis-4216 294 2 . . PUNCT fis-4216 295 1 constr constr PROPN fis-4216 295 2 . . PUNCT fis-4216 295 3 , , PUNCT fis-4216 295 4 133 133 NUM fis-4216 295 5 , , PUNCT fis-4216 295 6 103991 103991 NUM fis-4216 295 7 . . PUNCT fis-4216 296 1 doi doi NOUN fis-4216 296 2 : : PUNCT fis-4216 296 3 10.1016 10.1016 NUM fis-4216 296 4 / / SYM fis-4216 296 5 j.autcon j.autcon NOUN fis-4216 296 6 . . PUNCT fis-4216 297 1 2021.103991 2021.103991 NUM fis-4216 297 2 . . PUNCT fis-4216 298 1 [ [ X fis-4216 298 2 15 15 NUM fis-4216 298 3 ] ] X fis-4216 298 4 shang shang PROPN fis-4216 298 5 , , PUNCT fis-4216 298 6 j. j. PROPN fis-4216 298 7 , , PUNCT fis-4216 298 8 xu xu PROPN fis-4216 298 9 , , PUNCT fis-4216 298 10 j. j. PROPN fis-4216 298 11 , , PUNCT fis-4216 298 12 zhang zhang PROPN fis-4216 298 13 , , PUNCT fis-4216 298 14 a. a. PROPN fis-4216 298 15 a. a. PROPN fis-4216 298 16 , , PUNCT fis-4216 298 17 liu liu PROPN fis-4216 298 18 , , PUNCT fis-4216 298 19 y. y. PROPN fis-4216 298 20 , , PUNCT fis-4216 298 21 wang wang PROPN fis-4216 298 22 , , PUNCT fis-4216 298 23 k. k. PROPN fis-4216 298 24 c. c. PROPN fis-4216 298 25 , , PUNCT fis-4216 298 26 ren ren PROPN fis-4216 298 27 , , PUNCT fis-4216 298 28 d. d. PROPN fis-4216 298 29 , , PUNCT fis-4216 298 30 he he PRON fis-4216 298 31 , , PUNCT fis-4216 298 32 a. a. NOUN fis-4216 298 33 , , PUNCT fis-4216 298 34 ( ( PUNCT fis-4216 298 35 2023 2023 NUM fis-4216 298 36 ) ) PUNCT fis-4216 298 37 . . PUNCT fis-4216 299 1 automatic automatic ADJ fis-4216 299 2 pixel pixel ADJ fis-4216 299 3 - - PUNCT fis-4216 299 4 level level NOUN fis-4216 299 5 pavement pavement NOUN fis-4216 299 6 sealed seal VERB fis-4216 299 7 crack crack NOUN fis-4216 299 8 detection detection NOUN fis-4216 299 9 using use VERB fis-4216 299 10 multi multi ADJ fis-4216 299 11 - - ADJ fis-4216 299 12 fusion fusion ADJ fis-4216 299 13 u u ADJ fis-4216 299 14 - - ADJ fis-4216 299 15 net net ADJ fis-4216 299 16 network network NOUN fis-4216 299 17 . . PUNCT fis-4216 300 1 measurement measurement PROPN fis-4216 300 2 , , PUNCT fis-4216 300 3 112475 112475 NUM fis-4216 300 4 . . PUNCT fis-4216 301 1 doi doi NOUN fis-4216 301 2 : : PUNCT fis-4216 301 3 10.1016 10.1016 NUM fis-4216 301 4 / / SYM fis-4216 301 5 j.measurement.2023.112475 j.measurement.2023.112475 PROPN fis-4216 301 6 . . PUNCT fis-4216 302 1 [ [ X fis-4216 302 2 16 16 NUM fis-4216 302 3 ] ] X fis-4216 302 4 ha ha INTJ fis-4216 302 5 , , PUNCT fis-4216 302 6 j. j. PROPN fis-4216 302 7 , , PUNCT fis-4216 302 8 kim kim PROPN fis-4216 302 9 , , PUNCT fis-4216 302 10 d. d. PROPN fis-4216 302 11 , , PUNCT fis-4216 302 12 kim kim PROPN fis-4216 302 13 , , PUNCT fis-4216 302 14 m. m. NOUN fis-4216 302 15 , , PUNCT fis-4216 302 16 ( ( PUNCT fis-4216 302 17 2022 2022 NUM fis-4216 302 18 ) ) PUNCT fis-4216 302 19 . . PUNCT fis-4216 303 1 assessing assess VERB fis-4216 303 2 severity severity NOUN fis-4216 303 3 of of ADP fis-4216 303 4 road road NOUN fis-4216 303 5 cracks crack NOUN fis-4216 303 6 using use VERB fis-4216 303 7 deep deep ADJ fis-4216 303 8 learning learning NOUN fis-4216 303 9 - - PUNCT fis-4216 303 10 based base VERB fis-4216 303 11 segmentation segmentation NOUN fis-4216 303 12 and and CCONJ fis-4216 303 13 detection detection NOUN fis-4216 303 14 . . PUNCT fis-4216 304 1 j. j. PROPN fis-4216 304 2 supercomput supercomput PROPN fis-4216 304 3 . . PUNCT fis-4216 304 4 , , PUNCT fis-4216 304 5 78(16 78(16 NOUN fis-4216 304 6 ) ) PUNCT fis-4216 304 7 , , PUNCT fis-4216 304 8 pp pp PROPN fis-4216 304 9 . . PUNCT fis-4216 305 1 17721 17721 NUM fis-4216 305 2 - - SYM fis-4216 305 3 17735 17735 NUM fis-4216 305 4 . . PUNCT fis-4216 306 1 doi doi NOUN fis-4216 306 2 : : PUNCT fis-4216 306 3 10.1007 10.1007 NUM fis-4216 306 4 / / SYM fis-4216 306 5 s11227 s11227 VERB fis-4216 306 6 - - PUNCT fis-4216 306 7 022 022 NUM fis-4216 306 8 - - PUNCT fis-4216 306 9 04560 04560 NUM fis-4216 306 10 - - PUNCT fis-4216 306 11 x. x. NOUN fis-4216 307 1 [ [ X fis-4216 307 2 17 17 NUM fis-4216 307 3 ] ] X fis-4216 307 4 ucar ucar PROPN fis-4216 307 5 , , PUNCT fis-4216 307 6 f. f. PROPN fis-4216 307 7 , , PUNCT fis-4216 307 8 korkmaz korkmaz PROPN fis-4216 307 9 , , PUNCT fis-4216 307 10 d. d. PROPN fis-4216 307 11 , , PUNCT fis-4216 307 12 ( ( PUNCT fis-4216 307 13 2020 2020 NUM fis-4216 307 14 ) ) PUNCT fis-4216 307 15 . . PUNCT fis-4216 308 1 covidiagnosis covidiagnosis NOUN fis-4216 308 2 - - PUNCT fis-4216 308 3 net net NOUN fis-4216 308 4 : : PUNCT fis-4216 308 5 deep deep ADJ fis-4216 308 6 bayes bayes NOUN fis-4216 308 7 - - PUNCT fis-4216 308 8 squeezenet squeezenet NOUN fis-4216 308 9 based base VERB fis-4216 308 10 diagnosis diagnosis NOUN fis-4216 308 11 of of ADP fis-4216 308 12 the the DET fis-4216 308 13 coronavirus coronavirus NOUN fis-4216 308 14 disease disease NOUN fis-4216 308 15 2019 2019 NUM fis-4216 308 16 ( ( PUNCT fis-4216 308 17 covid-19 covid-19 PROPN fis-4216 308 18 ) ) PUNCT fis-4216 308 19 from from ADP fis-4216 308 20 x x ADJ fis-4216 308 21 - - NOUN fis-4216 308 22 ray ray NOUN fis-4216 308 23 images image NOUN fis-4216 308 24 . . PUNCT fis-4216 309 1 med med ADJ fis-4216 309 2 . . PUNCT fis-4216 309 3 hypotheses hypothesis NOUN fis-4216 309 4 , , PUNCT fis-4216 309 5 140 140 NUM fis-4216 309 6 , , PUNCT fis-4216 309 7 109761 109761 NUM fis-4216 309 8 . . PUNCT fis-4216 310 1 doi doi NOUN fis-4216 310 2 : : PUNCT fis-4216 310 3 10.1016 10.1016 NUM fis-4216 310 4 / / SYM fis-4216 310 5 j.mehy.2020.109761 j.mehy.2020.109761 NOUN fis-4216 310 6 . . PUNCT fis-4216 311 1 [ [ X fis-4216 311 2 18 18 NUM fis-4216 311 3 ] ] PUNCT fis-4216 311 4 raghu raghu PROPN fis-4216 311 5 , , PUNCT fis-4216 311 6 s. s. PROPN fis-4216 311 7 , , PUNCT fis-4216 311 8 sriraam sriraam NOUN fis-4216 311 9 , , PUNCT fis-4216 311 10 n. n. NOUN fis-4216 311 11 , , PUNCT fis-4216 311 12 temel temel NOUN fis-4216 311 13 , , PUNCT fis-4216 311 14 y. y. PROPN fis-4216 311 15 , , PUNCT fis-4216 311 16 rao rao PROPN fis-4216 311 17 , , PUNCT fis-4216 311 18 s. s. PROPN fis-4216 311 19 v. v. PROPN fis-4216 311 20 , , PUNCT fis-4216 311 21 kubben kubben PROPN fis-4216 311 22 , , PUNCT fis-4216 311 23 p. p. PROPN fis-4216 311 24 l. l. PROPN fis-4216 311 25 , , PUNCT fis-4216 311 26 ( ( PUNCT fis-4216 311 27 2020 2020 NUM fis-4216 311 28 ) ) PUNCT fis-4216 311 29 . . PUNCT fis-4216 311 30 eeg eeg PROPN fis-4216 311 31 based base VERB fis-4216 311 32 multi multi ADJ fis-4216 311 33 - - ADJ fis-4216 311 34 class class ADJ fis-4216 311 35 seizure seizure NOUN fis-4216 311 36 type type NOUN fis-4216 311 37 classification classification NOUN fis-4216 311 38 using use VERB fis-4216 311 39 convolutional convolutional ADJ fis-4216 311 40 neural neural ADJ fis-4216 311 41 network network NOUN fis-4216 311 42 and and CCONJ fis-4216 311 43 transfer transfer NOUN fis-4216 311 44 learning learning NOUN fis-4216 311 45 . . PUNCT fis-4216 312 1 neural neural ADJ fis-4216 312 2 netw netw NOUN fis-4216 312 3 . . PUNCT fis-4216 313 1 , , PUNCT fis-4216 313 2 124 124 NUM fis-4216 313 3 , , PUNCT fis-4216 313 4 202 202 NUM fis-4216 313 5 - - SYM fis-4216 313 6 212 212 NUM fis-4216 313 7 . . PUNCT fis-4216 314 1 doi doi NOUN fis-4216 314 2 : : PUNCT fis-4216 314 3 10.1016 10.1016 NUM fis-4216 314 4 / / SYM fis-4216 314 5 j. j. PROPN fis-4216 314 6 neunet neunet PROPN fis-4216 314 7 . . PUNCT fis-4216 315 1 2020.01.017 2020.01.017 X fis-4216 315 2 . . PUNCT fis-4216 316 1 [ [ X fis-4216 316 2 19 19 NUM fis-4216 316 3 ] ] X fis-4216 316 4 buda buda NOUN fis-4216 316 5 , , PUNCT fis-4216 316 6 m. m. NOUN fis-4216 316 7 , , PUNCT fis-4216 316 8 maki maki PROPN fis-4216 316 9 , , PUNCT fis-4216 316 10 a. a. NOUN fis-4216 316 11 , , PUNCT fis-4216 316 12 mazurowski mazurowski VERB fis-4216 316 13 , , PUNCT fis-4216 316 14 m. m. NOUN fis-4216 316 15 a. a. PROPN fis-4216 316 16 , , PUNCT fis-4216 316 17 ( ( PUNCT fis-4216 316 18 2018 2018 NUM fis-4216 316 19 ) ) PUNCT fis-4216 316 20 . . PUNCT fis-4216 317 1 a a DET fis-4216 317 2 systematic systematic ADJ fis-4216 317 3 study study NOUN fis-4216 317 4 of of ADP fis-4216 317 5 the the DET fis-4216 317 6 class class NOUN fis-4216 317 7 imbalance imbalance NOUN fis-4216 317 8 problem problem NOUN fis-4216 317 9 in in ADP fis-4216 317 10 convolutional convolutional ADJ fis-4216 317 11 neural neural ADJ fis-4216 317 12 networks network NOUN fis-4216 317 13 . . PUNCT fis-4216 318 1 neural neural ADJ fis-4216 318 2 netw netw NOUN fis-4216 318 3 . . PUNCT fis-4216 318 4 , , PUNCT fis-4216 318 5 106 106 NUM fis-4216 318 6 , , PUNCT fis-4216 318 7 pp pp ADJ fis-4216 318 8 . . PUNCT fis-4216 319 1 249 249 NUM fis-4216 319 2 - - SYM fis-4216 319 3 259 259 NUM fis-4216 319 4 . . PUNCT fis-4216 319 5 doi doi NOUN fis-4216 319 6 : : PUNCT fis-4216 319 7 10.1016 10.1016 NUM fis-4216 319 8 / / SYM fis-4216 319 9 j.neunet.2018.07.011 j.neunet.2018.07.011 PROPN fis-4216 319 10 . . PUNCT fis-4216 320 1 [ [ X fis-4216 320 2 20 20 NUM fis-4216 320 3 ] ] PUNCT fis-4216 320 4 iandola iandola PROPN fis-4216 320 5 , , PUNCT fis-4216 320 6 f. f. PROPN fis-4216 320 7 n. n. PROPN fis-4216 320 8 , , PUNCT fis-4216 320 9 han han PROPN fis-4216 320 10 , , PUNCT fis-4216 320 11 s. s. PROPN fis-4216 320 12 , , PUNCT fis-4216 320 13 moskewicz moskewicz NOUN fis-4216 320 14 , , PUNCT fis-4216 320 15 m. m. NOUN fis-4216 320 16 w. w. PROPN fis-4216 320 17 , , PUNCT fis-4216 320 18 ashraf ashraf PROPN fis-4216 320 19 , , PUNCT fis-4216 320 20 k. k. PROPN fis-4216 320 21 , , PUNCT fis-4216 320 22 dally dally ADV fis-4216 320 23 , , PUNCT fis-4216 320 24 w. w. PROPN fis-4216 320 25 j. j. PROPN fis-4216 320 26 , , PUNCT fis-4216 320 27 keutzer keutzer PROPN fis-4216 320 28 , , PUNCT fis-4216 320 29 k. k. PROPN fis-4216 320 30 , , PUNCT fis-4216 320 31 ( ( PUNCT fis-4216 320 32 2016 2016 NUM fis-4216 320 33 ) ) PUNCT fis-4216 320 34 . . PUNCT fis-4216 321 1 squeezenet squeezenet NOUN fis-4216 321 2 : : PUNCT fis-4216 321 3 alexnet alexnet ADJ fis-4216 321 4 - - PUNCT fis-4216 321 5 level level NOUN fis-4216 321 6 accuracy accuracy NOUN fis-4216 321 7 with with ADP fis-4216 321 8 50x 50x NUM fis-4216 321 9 fewer few ADJ fis-4216 321 10 parameters parameter NOUN fis-4216 321 11 and and CCONJ fis-4216 321 12 < < X fis-4216 321 13 0.5 0.5 NUM fis-4216 321 14 mb mb NOUN fis-4216 321 15 model model NOUN fis-4216 321 16 size size NOUN fis-4216 321 17 . . PUNCT fis-4216 322 1 arxiv arxiv PROPN fis-4216 322 2 preprint preprint NOUN fis-4216 322 3 arxiv:1602.07360 arxiv:1602.07360 NOUN fis-4216 322 4 . . PUNCT fis-4216 323 1 doi doi NOUN fis-4216 323 2 : : PUNCT fis-4216 323 3 10.48550 10.48550 NUM fis-4216 323 4 / / SYM fis-4216 323 5 arxiv.1602.07360 arxiv.1602.07360 NOUN fis-4216 323 6 . . PUNCT fis-4216 324 1 [ [ X fis-4216 324 2 21 21 NUM fis-4216 324 3 ] ] X fis-4216 324 4 kenta kenta PROPN fis-4216 324 5 , , PUNCT fis-4216 324 6 ( ( PUNCT fis-4216 324 7 2023 2023 NUM fis-4216 324 8 ) ) PUNCT fis-4216 324 9 . . PUNCT fis-4216 325 1 classify classify VERB fis-4216 325 2 crack crack VERB fis-4216 325 3 image image NOUN fis-4216 325 4 using use VERB fis-4216 325 5 deep deep ADJ fis-4216 325 6 learning learning NOUN fis-4216 325 7 and and CCONJ fis-4216 325 8 explain explain VERB fis-4216 325 9 " " PUNCT fis-4216 325 10 why why SCONJ fis-4216 325 11 " " PUNCT fis-4216 325 12 . . PUNCT fis-4216 326 1 https://github.com https://github.com X fis-4216 326 2 /kentaitakura /kentaitakura SYM fis-4216 326 3 / / SYM fis-4216 326 4 releases release NOUN fis-4216 326 5 / / SYM fis-4216 326 6 tag tag NOUN fis-4216 326 7 / / SYM fis-4216 326 8 v1.1 v1.1 NOUN fis-4216 326 9 [ [ X fis-4216 326 10 22 22 NUM fis-4216 326 11 ] ] X fis-4216 326 12 long long ADV fis-4216 326 13 , , PUNCT fis-4216 326 14 j. j. PROPN fis-4216 326 15 , , PUNCT fis-4216 326 16 shelhamer shelhamer PROPN fis-4216 326 17 , , PUNCT fis-4216 326 18 e. e. PROPN fis-4216 326 19 , , PUNCT fis-4216 326 20 darrell darrell PROPN fis-4216 326 21 , , PUNCT fis-4216 326 22 t. t. PROPN fis-4216 326 23 , , PUNCT fis-4216 326 24 ( ( PUNCT fis-4216 326 25 2015 2015 NUM fis-4216 326 26 ) ) PUNCT fis-4216 326 27 . . PUNCT fis-4216 327 1 fully fully ADV fis-4216 327 2 convolutional convolutional ADJ fis-4216 327 3 networks network NOUN fis-4216 327 4 for for ADP fis-4216 327 5 semantic semantic ADJ fis-4216 327 6 segmentation segmentation NOUN fis-4216 327 7 . . PUNCT fis-4216 328 1 in in ADP fis-4216 328 2 proceedings proceeding NOUN fis-4216 328 3 of of ADP fis-4216 328 4 the the DET fis-4216 328 5 ieee ieee NOUN fis-4216 328 6 conference conference NOUN fis-4216 328 7 on on ADP fis-4216 328 8 computer computer NOUN fis-4216 328 9 vision vision NOUN fis-4216 328 10 and and CCONJ fis-4216 328 11 pattern pattern NOUN fis-4216 328 12 recognition recognition NOUN fis-4216 328 13 , , PUNCT fis-4216 328 14 pp pp ADP fis-4216 328 15 . . PUNCT fis-4216 328 16 3431 3431 NUM fis-4216 328 17 - - SYM fis-4216 328 18 3440 3440 NUM fis-4216 328 19 . . PUNCT fis-4216 329 1 doi doi NOUN fis-4216 329 2 : : PUNCT fis-4216 329 3 10.48550 10.48550 NUM fis-4216 329 4 / / SYM fis-4216 329 5 arxiv.1411.4038 arxiv.1411.4038 PROPN fis-4216 329 6 . . PUNCT fis-4216 330 1 https://doi.org/10.1016/j.neunet.2018.07.011 https://doi.org/10.1016/j.neunet.2018.07.011 PROPN fis-4216 330 2 https://doi.org/10.48550/arxiv.1411.4038 https://doi.org/10.48550/arxiv.1411.4038 PROPN