Research map / Label-efficient and robust learning / Semi-supervised learning
Unbiased Subclass Regularization for Semi-Supervised Semantic Segmentation
Dayan Guan, Jiaxing Huang, Aoran Xiao, Shijian Lu
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add class-unbiased subclass regularization (ablation) | Original-class self-training on USRN · PASCAL VOC 1/32 | mIoU: 64.1 -> 65.0 | Improved |
| Add subclass self-training (ablation) | USRN with MSL, OST, and USR · PASCAL VOC 1/32 | mIoU: +2.1% in mIoU | Improved |
| Add the entropy-based gate mechanism (ablation) | USRN with subclass self-training · PASCAL VOC 1/32 | mIoU: +1.5% in mIoU | Improved |
| Replace balanced k-means with normal k-means (ablation) | USRN · PASCAL VOC 1/32 | mIoU: 68.6 -> 68.0 | Hurt |
| Replace balanced k-means with normal k-means (ablation) | Subclass annotations · PASCAL VOC 1/32 | Subclass CBR: 99.5% -> 96.4% | Hurt |
| Remove feature sharing between F and F_sub (ablation) | USRN · PASCAL VOC 1/32 | mIoU: 68.6 -> 67.3 | Hurt |
| Share both low- and high-level features (ablation) | USRN · PASCAL VOC 1/32 | mIoU: 68.6 -> 67.8 | Hurt |
| Set pseudo-label confidence threshold γ to 0.85 or 0.95 (ablation) | USRN · PASCAL VOC 1/32 | mIoU: 68.6; 68.5 | No clear effect |
| Set pseudo-label confidence threshold γ below 0.75 (ablation) | USRN · PASCAL VOC 1/32 | mIoU: 68.6 -> 67.7; 67.4 | Hurt |
| Set pseudo-label confidence threshold γ above 0.95 (ablation) | USRN · PASCAL VOC 1/32 | mIoU: 68.6 -> 68.1 | Hurt |
| Use USRN instead of the strongest reported baseline | Semi-supervised segmentation · PASCAL VOC 1/64; Cityscapes 1/32 | mIoU: 3.8%; 2.1% | Improved |
| Integrate USRN into existing semi-supervised methods | DARS, CPS, and CAC · PASCAL VOC 1/32 | mIoU: +4.5; +4.4; +4.9 | Improved |
Closest papers by meaning
- Towards Unbiased Learning in Semi-Supervised Semantic Segmentation (ICLR 2025 · similarity 0.90)
- Regularized Loss for Weakly Supervised Single Class Semantic Segmentation (ECCV 2020 · similarity 0.89)
- DuPL: Dual Student with Trustworthy Progressive Learning for Robust Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.88)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.88)
- Relaxed Class-consensus Consistency for Semi-supervised Semantic Segmentation (AAAI 2025 · similarity 0.88)
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
