Research map / Detection and segmentation / Weakly supervised detection and segmentation
Group-Wise Semantic Mining for Weakly 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 |
|---|---|---|---|
| Use group-wise graph reasoning over image groups | Group-WSSS · PASCAL VOC 2012 val | mIoU: 67.8% -> 68.2% | Helped |
| Use group-wise graph reasoning over image groups | Group-WSSS · COCO val | mIoU: 27.7% -> 28.4% | Helped |
| Use K=3 images instead of K=4 (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 68.2% -> 68.1% | No clear effect |
| Use more than four images per group (K=5 or K=6) (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 68.2% -> 67.8%; 68.2% -> 67.6% | Hurt |
| Use T=2 message-passing steps instead of T=3 (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 67.8% -> 68.2% | Hurt |
| Use more than three message-passing steps (T=4 or T=5) (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 68.2% -> 68.0% | Hurt |
| Set graph-dropout δd to 0.9 or 0.5 instead of 0.7 (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 68.2% -> 68.0%; 68.2% -> 67.7% | Hurt |
| Lower graph-dropout δr to 0.6 or 0.4 (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 68.2% -> 66.8%; 68.2% -> 63.6% | Hurt |
| Add graph dropout to suppress highly activated regions (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 67.7% -> 68.2% | Helped |
| Use graph readout instead of intermediate readout only (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 64.1% -> 67.8% | Helped |
| Ensemble intermediate and graph outputs (ablation) | Group-WSSS · PASCAL VOC 2012 val | mIoU: 67.8% -> 68.2% | Helped |
Closest papers by meaning
- Uncovering Prototypical Knowledge for Weakly Open-Vocabulary Semantic Segmentation (NeurIPS 2023 · similarity 0.90)
- Webly Supervised Semantic Segmentation (CVPR 2017 · similarity 0.89)
- Separate and Conquer: Decoupling Co-occurrence via Decomposition and Representation for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.88)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.87)
- From SAM to CAMs: Exploring Segment Anything Model for Weakly Supervised Semantic Segmentation (CVPR 2024 · 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.