Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Semantic Segmentation via Gentle Teaching Assistant
Read the paperCheck 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 |
|---|---|---|---|
| add teacher-student framework (ablation) | SupOnly · VOC original 183 | mIoU: 54.92 -> 58.93 | Helped |
| add gentle teaching assistant (ablation) | teacher-student · VOC original 183 | mIoU: 58.93 -> 72.10 | Helped |
| add re-weighted pseudo labeling (ablation) | teacher-student + GTA · VOC original 183 | mIoU: 72.10 -> 73.16 | Helped |
| transmit only encoder via EMA instead of all parameters (ablation) | GTA transmission · VOC original 183 | mIoU: 64.07 -> 72.10 | Helped |
| transmit encoder via EMA instead of Unbiased ST (ablation) | GTA transmission · VOC original 183 | mIoU: 65.92 -> 72.10 | Helped |
| train student on pseudo labels and GTA on labeled data (vs reverse) (ablation) | GTA-Seg · VOC original 183 | mIoU: 73.16 -> 66.71 | Hurt |
| add labeled data to GTA training (ablation) | GTA-Seg · VOC original 183 | mIoU: 73.16 -> 72.28 | No clear effect |
| add confidence-based re-weighting without Laplace smoothing (ablation) | GTA-Seg · VOC original 183 | mIoU: 72.10 -> 70.67 | Hurt |
| add Laplace smoothing to confidence re-weighting (ablation) | GTA-Seg · VOC original 183 | mIoU: 70.67 -> 73.16 | Helped |
| increase EMA alpha from 0.99 to 0.9999 (ablation) | GTA-Seg · VOC original 183 | mIoU: 73.16 -> 73.57 | No clear effect |
| increase warmup epochs from 1 to 2 or 3 (ablation) | GTA-Seg · VOC original 183 | mIoU: 73.16 -> 73.58 | No clear effect |
Closest papers by meaning
- Learning from Future: A Novel Self-Training Framework for Semantic Segmentation (NeurIPS 2022 · similarity 0.92)
- Switching Temporary Teachers for Semi-Supervised Semantic Segmentation (NeurIPS 2023 · similarity 0.92)
- PseudoSeg: Designing Pseudo Labels for Semantic Segmentation (ICLR 2021 · similarity 0.91)
- ST++: Make Self-Training Work Better for Semi-Supervised Semantic Segmentation (CVPR 2022 · similarity 0.90)
- Instance-Specific and Model-Adaptive Supervision for Semi-Supervised Semantic Segmentation (CVPR 2023 · similarity 0.90)
- AllSpark: Reborn Labeled Features from Unlabeled in Transformer for Semi-Supervised Semantic Segmentation (CVPR 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.