Research map / Detection and segmentation / Weakly supervised detection and segmentation
Revisiting Weak-to-Strong Consistency in Semi-Supervised Semantic Segmentation
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 feature-level channel dropout to FixMatch (ablation) | FixMatch on Pascal · Pascal, 92 labeled images | mIoU: 63.9 -> 72.0 | Helped |
| use two strong image views guided by a shared weak view (ablation) | FixMatch on Pascal · Pascal, 92 labeled images | mIoU: 63.9 -> 72.1 | Helped |
| combine feature-level perturbation and dual strong image views | FixMatch on Pascal · Pascal, 92 labeled images | mIoU: 63.9 -> 75.2 | Helped |
| replace the feature-level view with a third strong image view (ablation) | UniMatch on Pascal · Pascal, 92 labeled images | mIoU: 75.2 -> 71.6 | Hurt |
| double unlabeled batch size instead of using dual strong views (ablation) | FixMatch on Pascal · Pascal, 92 labeled images | mIoU: 72.1 vs. 62.5 | Hurt |
| double training epochs instead of using dual strong views (ablation) | FixMatch on Pascal · Pascal, 92 labeled images | mIoU: 72.1 vs. 61.8 | Hurt |
| apply feature dropout to a strongly perturbed image stream rather than separating perturbation levels (ablation) | UniPerb on Pascal · Pascal, 92 labeled images | mIoU: 72.0 -> 63.4 | Hurt |
| use feature-level perturbation alone instead of image-level strong perturbation (ablation) | FixMatch on Pascal · Pascal, 92 labeled images | mIoU: 63.9 -> 66.0; 73.0 -> 69.6; 75.5 -> 74.0; 77.8 -> 77.3; 79.2 -> 78.9 | Mixed |
| add a second feature-level dropout stream (ablation) | UniPerb on Pascal · Pascal, 92 labeled images | mIoU: 72.0 -> 73.4 | Helped |
| move feature perturbation to the decoder-classifier input (ablation) | UniPerb on Pascal · Pascal, 92 labeled images | mIoU: 72.0 -> 64.5 | Hurt |
| double the number of image- and feature-level perturbation streams (ablation) | UniMatch on Pascal · Pascal, 183 labeled images | mIoU: 77.2 -> 77.7 | Mixed |
Closest papers by meaning
- Perturbed and Strict Mean Teachers for Semi-Supervised Semantic Segmentation (CVPR 2022 · similarity 0.89)
- Towards the Uncharted: Density-Descending Feature Perturbation for Semi-supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Semi-Supervised Semantic Segmentation With Cross-Consistency Training (CVPR 2020 · similarity 0.88)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.88)
- ScaleMatch: Multi-scale Consistency Enhancement for Semi-supervised Semantic Segmentation (AAAI 2025 · similarity 0.88)
- GapMatch: Bridging Instance and Model Perturbations for Enhanced Semi-Supervised Medical Image Segmentation (AAAI 2025 · 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.