Unbiased Subclass Regularization for Semi-Supervised Semantic Segmentation

被引:21
|
作者
Guan, Dayan [1 ]
Huang, Jiaxing [1 ]
Xiao, Aoran [1 ]
Lu, Shijian [1 ]
机构
[1] Nanyang Technol Univ, Singtel Cognit & Artificial Intelligence Lab Ente, Singapore, Singapore
关键词
D O I
10.1109/CVPR52688.2022.00973
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Semi-supervised semantic segmentation learns from small amounts of labelled images and large amounts of unlabelled images, which has witnessed impressive progress with the recent advance of deep neural networks. However, it often suffers from severe class-bias problem while exploring the unlabelled images, largely due to the clear pixel-wise class imbalance in the labelled images. This paper presents an unbiased subclass regularization network (USRN) that alleviates the class imbalance issue by learning class-unbiased segmentation from balanced subclass distributions. We build the balanced subclass distributions by clustering pixels of each original class into multiple subclasses of similar sizes, which provide class-balanced pseudo supervision to regularize the class-biased segmentation. In addition, we design an entropy-based gate mechanism to coordinate learning between the original classes and the clustered subclasses which facilitates subclass regularization effectively by suppressing unconfident subclass predictions. Extensive experiments over multiple public benchmarks show that USRN achieves superior performance as compared with the state-of-the-art.
引用
收藏
页码:9958 / 9968
页数:11
相关论文
共 50 条
  • [41] Adversarial Dense Contrastive Learning for Semi-Supervised Semantic Segmentation
    Wang, Ying
    Xuan, Ziwei
    Ho, Chiuman
    Qi, Guo-Jun
    IEEE TRANSACTIONS ON IMAGE PROCESSING, 2023, 32 : 4459 - 4471
  • [42] Semi-supervised Dual-Domain Adaptation for Semantic Segmentation
    Chen, Ying
    Xu Ouyang
    Zhu, Kaiyue
    Agam, Gady
    2022 26TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR), 2022, : 230 - 237
  • [43] SEMI-SUPERVISED SEMANTIC SEGMENTATION OF HYPER-SPECTRAL IMAGES
    Pal, Shankho Subhra
    Mukhopadhyay, Jayanta
    Sarkar, Sudeshna
    IGARSS 2023 - 2023 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2023, : 7657 - 7660
  • [44] Semi-supervised Semantic Segmentation of Multiple Lumbosacral Structures on CT
    Liu, Huaqing
    Xiao, Haoping
    Luo, Lishu
    Feng, Chaobo
    Yin, Bangde
    Wang, Dongdong
    Li, Yufeng
    He, Shisheng
    Fan, Guoxin
    COMPUTATIONAL METHODS AND CLINICAL APPLICATIONS FOR SPINE IMAGING, CSI 2019, 2020, 11963 : 47 - 59
  • [45] Semi-Supervised Learning for Semantic Segmentation of Emphysema With Partial Annotations
    Peng, Liying
    Lin, Lanfen
    Hu, Hongjie
    Zhang, Yue
    Li, Huali
    Iwamoto, Yutaro
    Han, Xian-Hua
    Chen, Yen-Wei
    IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, 2020, 24 (08) : 2327 - 2336
  • [46] Linear semantic transformation for semi-supervised medical image segmentation
    Chen C.
    Chen Y.
    Li X.
    Ning H.
    Xiao R.
    Computers in Biology and Medicine, 2024, 173
  • [47] Casting Geometric Constraints in Semantic Segmentation as Semi-Supervised Learning
    Stekovic, Sinisa
    Fraundorfer, Friedrich
    Lepetit, Vincent
    2020 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV), 2020, : 1843 - 1852
  • [48] Perturbed and Strict Mean Teachers for Semi-supervised Semantic Segmentation
    Liu, Yuyuan
    Tian, Yu
    Chen, Yuanhong
    Liu, Fengbei
    Belagiannis, Vasileios
    Carneiro, Gustavo
    2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022), 2022, : 4248 - 4257
  • [49] Semi-supervised Semantic Segmentation via Prototypical Contrastive Learning
    Chen, Zenggui
    Lian, Zhouhui
    PROCEEDINGS OF THE 30TH ACM INTERNATIONAL CONFERENCE ON MULTIMEDIA, MM 2022, 2022, : 6696 - 6705
  • [50] Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation
    Hong, Seunghoon
    Noh, Hyeonwoo
    Han, Bohyung
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 28 (NIPS 2015), 2015, 28