Research map / Label-efficient and robust learning / Semi-supervised learning
Boosting Semi-Supervised Semantic Segmentation with Probabilistic Representations
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 |
|---|---|---|---|
| Replace deterministic representations and point prototypes with probabilistic representations and distribution prototypes (ablation) | PRCL · Pascal VOC, 60 labels, δ_s=0.80, δ_w=0.70 | mIoU: 55.45 -> 64.27 | Helped |
| Replace deterministic representations and point prototypes with probabilistic representations and distribution prototypes (ablation) | PRCL · Pascal VOC, 92 labels, δ_s=0.80, δ_w=0.70 | mIoU: 62.70 -> 67.91 | Helped |
| Lower the anchor confidence threshold δ_s from 0.97 to 0.80 (ablation) | PRCL with probabilistic representations · Pascal VOC, 60 labels | mIoU: 61.11 -> 64.27 | Helped |
| Lower the anchor confidence threshold δ_s from 0.97 to 0.80 (ablation) | PRCL without probabilistic mechanism · Pascal VOC, 60 labels | mIoU: 61.83 -> 55.45 | Hurt |
| Remove the sampling strategy (δ_s=1.00, δ_w=0.00) (ablation) | PRCL with probabilistic representations · Pascal VOC, 60 labels | mIoU: 61.74 -> 56.78 | Hurt |
| Use a contrastive-loss weight scheduler (ablation) | PRCL · Pascal VOC, 46 labels | mIoU: 43.31 -> 43.00 | Hurt |
| Use a contrastive-loss weight scheduler (ablation) | PRCL · Pascal VOC, 60 labels | mIoU: 64.27 -> 58.10 | Hurt |
| Use a contrastive-loss weight scheduler (ablation) | PRCL · Pascal VOC, 92 labels | mIoU: 67.91 -> 68.52 | Helped |
| Use a contrastive-loss weight scheduler (ablation) | PRCL · Pascal VOC, 183 labels | mIoU: 69.40 -> 72.14 | Helped |
| Increase the number of sampled negatives (ablation) | PRCL · Pascal VOC | mIoU: 65.53 -> 65.72 -> 67.53 -> 67.91 | Helped |
| Use probabilistic representations to tolerate uncertain pseudo-labels rather than relying only on deterministic features | PRCL · Pascal VOC and CityScapes | mIoU | Helped |
Closest papers by meaning
- Semi-Supervised Semantic Segmentation With Directional Context-Aware Consistency (CVPR 2021 · similarity 0.91)
- Looking Beyond Single Images for Contrastive Semantic Segmentation Learning (NeurIPS 2021 · similarity 0.90)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.87)
- RankMatch: Exploring the Better Consistency Regularization for Semi-supervised Semantic Segmentation (CVPR 2024 · similarity 0.87)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.87)
- 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.