Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Semantic Segmentation With Cross Pseudo Supervision
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use cross pseudo supervision with two differently initialized networks instead of supervised training alone | DeepLabv3+ on Cityscapes · Cityscapes val, 1/16 labeled, ResNet-50 | mIoU: +4.89% | Helped |
| add CutMix augmentation (ablation) | CPS on Cityscapes · Cityscapes val, 1/16 labeled, ResNet-101 | mIoU: +4.22% | Helped |
| add CutMix augmentation (ablation) | CPS on PASCAL VOC 2012 · PASCAL VOC 2012 val, 1/16 labeled, ResNet-50 | mIoU: 3.77% | Helped |
| add CPS loss on unlabeled data (ablation) | supervised baseline · PASCAL VOC 2012 val, 1/8 labeled, ResNet-50 | mIoU: 69.43 -> 73.00 | Helped |
| add CPS loss on labeled data (ablation) | supervised baseline · PASCAL VOC 2012 val, 1/8 labeled, ResNet-50 | mIoU: 69.43 -> 69.99 | Helped |
| add CPS loss on both labeled and unlabeled data (ablation) | supervised baseline · Cityscapes val, 1/8 labeled, ResNet-50 | mIoU: 70.32 -> 74.39 | Helped |
| replace cross pseudo supervision with cross probability consistency (ablation) | two-network consistency training · Cityscapes val, 1/8 labeled, ResNet-50 | mIoU: 74.39 vs 72.03 | Hurt |
| replace two independently initialized networks with a single network (ablation) | pseudo supervision · PASCAL VOC 2012 val, 1/16 labeled | mIoU: 59.54 vs 68.21 | Hurt |
| add conventional self-training to CPS (ablation) | CPS on Cityscapes · Cityscapes val, 1/4 labeled, ResNet-50 | mIoU: 76.85 -> 77.40 | Helped |
| increase the trade-off weight λ (ablation) | CPS · PASCAL VOC 2012 and Cityscapes, 1/8 partition | mIoU | Mixed |
Closest papers by meaning
- Semi-Supervised Semantic Segmentation With Cross-Consistency Training (CVPR 2020 · similarity 0.94)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.89)
- RankMatch: Exploring the Better Consistency Regularization for Semi-supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
- ScaleMatch: Multi-scale Consistency Enhancement for Semi-supervised Semantic Segmentation (AAAI 2025 · similarity 0.88)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.88)
- RWMS: Reliable Weighted Multi-Phase for Semi-supervised Segmentation (AAAI 2024 · similarity 0.88)
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.