id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
ajst-11366	Che, Xuan; Zhu, Wenzhong	Tile Surface Defect Detection Based on Improved Faster R-CNN	2023	7	.pdf	application/pdf	3742	218	45	These results indicate that the improved network in this study performs well in tile defect detection, effectively enhancing the network's capability to detect small target defects. Although the incorporation of ResNet and the more complex FPN increases the number of parameters, the use of depthwise separable convolution reduces detection time, still meeting the requirements for tile defect detection on factory production lines.	cache/ajst-11366.pdf	txt/ajst-11366.txt