中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 计算技术研究所
  3. 中国科学院计算技术研究所
  4. 中国科学院计算技术研究所期刊论文
  5. 英文
BLPSeg: Balance the Label Preference in Scribble-Supervised Semantic Segmentation

文献类型：期刊论文

作者Wang, Yude2; Zhang, Jie2; Kan, Meina2; Shan, Shiguang1,2,3; Chen, Xilin2
刊名IEEE TRANSACTIONS ON IMAGE PROCESSING
出版日期2023
卷号32页码:4921-4934
关键词Annotations Semantic segmentation Training Task analysis Head Semantics Costs Scribble-supervised weakly supervised semantic segmentation
ISSN号1057-7149
DOI10.1109/TIP.2023.3301342
英文摘要Scribble-supervised semantic segmentation is an appealing weakly supervised technique with low labeling cost. Existing approaches mainly consider diffusing the labeled region of scribble by low-level feature similarity to narrow the supervision gap between scribble labels and mask labels. In this study, we observe an annotation bias between scribble and object mask, i.e., label workers tend to scribble on the spacious region instead of corners. This label preference makes the model learn well on those frequently labeled regions but poor on rarely labeled pixels. Therefore, we propose BLPSeg to balance the label preference for complete segmentation. Specifically, the BLPSeg first predicts an annotation probability map to evaluate the rarity of labels on each image, then utilizes a novel BLP loss to balance the model training by up-weighting those rare annotations. Additionally, to further alleviate the impact of label preference, we design a local aggregation module (LAM) to propagate supervision from labeled to unlabeled regions in gradient backpropagation. We conduct extensive experiments to illustrate the effectiveness of our BLPSeg. Our single-stage method even outperforms other advanced multi-stage methods and achieves state-of-the-art performance.
资助项目National Key Research and Development Program of China[2021YFC3310100] ; National Natural Science Foundation of China[62176251] ; National Natural Science Foundation of China[62122074] ; National Natural Science Foundation of China[61976219]
WOS研究方向Computer Science ; Engineering
语种英语
WOS记录号WOS:001067700800002
出版者IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
源URL[http://119.78.100.204/handle/2XEOYT63/21151]  
专题中国科学院计算技术研究所期刊论文_英文
通讯作者Zhang, Jie
作者单位1.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
2.Chinese Acad Sci, Inst Comp Technol, Key Lab Intelligent Informat Proc, CAS, Beijing 100190, Peoples R China
3.Peng Cheng Natl Lab, Shenzhen 518055, Peoples R China
推荐引用方式
GB/T 7714		 Wang, Yude,Zhang, Jie,Kan, Meina,et al. BLPSeg: Balance the Label Preference in Scribble-Supervised Semantic Segmentation[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2023,32:4921-4934.
APA Wang, Yude,Zhang, Jie,Kan, Meina,Shan, Shiguang,&Chen, Xilin.(2023).BLPSeg: Balance the Label Preference in Scribble-Supervised Semantic Segmentation.IEEE TRANSACTIONS ON IMAGE PROCESSING,32,4921-4934.
MLA Wang, Yude,et al."BLPSeg: Balance the Label Preference in Scribble-Supervised Semantic Segmentation".IEEE TRANSACTIONS ON IMAGE PROCESSING 32(2023):4921-4934.

入库方式： OAI收割

来源：计算技术研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。