Research map / Detection and segmentation / Weakly supervised detection and segmentation
Self-Supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentation
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 |
|---|---|---|---|
| Add equivariant regularization (ablation) | Baseline CAM · PASCAL VOC 2012 train pseudo labels | mIoU: 47.43% -> 49.90% | Helped |
| Add pixel correlation module to equivariant regularization (ablation) | Baseline CAM + ER · PASCAL VOC 2012 train pseudo labels | mIoU: 49.90% -> 55.08% | Helped |
| Add OHEM to equivariant cross regularization loss (ablation) | Baseline CAM + ER + PCM · PASCAL VOC 2012 train pseudo labels | mIoU: 55.08% -> 55.41% | Helped |
| Add dense CRF post-processing (ablation) | Baseline CAM · PASCAL VOC 2012 train pseudo labels | mIoU: 47.43% -> 52.40% | Helped |
| Add dense CRF post-processing (ablation) | SEAM · PASCAL VOC 2012 train pseudo labels | mIoU: 55.41% -> 56.83% | Helped |
| Replace CAM with GradCAM localization (ablation) | Weakly supervised localization · PASCAL VOC 2012 train pseudo labels | mIoU: 47.43% -> 46.53% | Hurt |
| Replace CAM with GradCAM++ localization (ablation) | Weakly supervised localization · PASCAL VOC 2012 train pseudo labels | mIoU: 47.43% -> 47.37% | No clear effect |
| Add rescaling equivariance regularization (ablation) | Baseline CAM · PASCAL VOC 2012 train pseudo labels | mIoU: 47.43% -> 55.41% | Helped |
| Add horizontal flip to rescaling (ablation) | SEAM equivariant regularization · PASCAL VOC 2012 train pseudo labels | mIoU: 55.41% -> 55.50% | No clear effect |
| Add random rotation to rescaling (ablation) | SEAM equivariant regularization · PASCAL VOC 2012 train pseudo labels | mIoU: 55.41% -> 53.13% | Hurt |
| Add translation to rescaling (ablation) | SEAM equivariant regularization · PASCAL VOC 2012 train pseudo labels | mIoU: 55.41% -> 55.23% | No clear effect |
| Expand baseline augmentation rescaling range to [224, 768] (ablation) | Baseline CAM · PASCAL VOC 2012 train pseudo labels | mIoU: 47.43% -> 46.72% | Hurt |
Closest papers by meaning
- Activation Modulation and Recalibration Scheme for Weakly Supervised Semantic Segmentation (AAAI 2022 · similarity 0.88)
- Equivariant Self-Supervised Learning: Encouraging Equivariance in Representations (ICLR 2022 · similarity 0.88)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.86)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.85)
- From SAM to CAMs: Exploring Segment Anything Model for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.85)
- Class Tokens Infusion for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.85)
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.