id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
ajst-30179	Liu, Linlin	Research on Semantic Segmentation Algorithms for Street Scenes	2025	5	.pdf	application/pdf	2890	164	36	Semantic segmentation is better suited for global environmental understanding, such as autonomous driving (for drivable area segmentation) or medical imaging (for organ and lesion localization).Today, semantic segmentation has been widely deployed across multiple domains. Compared to traditional classification networks, their key innovation lies in adopting a fully convolutional design, where convolutional layers replace fully connected layers as the classification output, preserving the spatial dimension of feature maps.	cache/ajst-30179.pdf	txt/ajst-30179.txt