Research map / Detection and segmentation / Weakly supervised detection and segmentation
Semantic-Aware Superpixel 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 DINO self-supervised transformer features instead of RGB or CNN features for superpixel discovery (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 | superpixel quality | Helped |
| use superpixel threshold τ=0.3 rather than τ=0.5 (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 62.9 -> 65.1 | Helped |
| use superpixel threshold τ=0.3 rather than τ=0.5 (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU after CRF: 67.5 -> 69.5 | Helped |
| use region growing with superpixels rather than no region growing (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 53.6 -> 66.3 | Helped |
| use region growing with superpixels rather than no region growing (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU after CRF: 59.7 -> 69.4 | Helped |
| set initial-seed threshold α to 0.6 rather than 0.3 while using region growing (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 59.7 -> 66.3 | Helped |
| set initial-seed threshold α to 0.7 rather than 0.6 while using region growing (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 66.3 -> 66.7 | Helped |
| increase α from 0.7 to 0.8 while using region growing (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 66.7 -> 65.5 | Hurt |
| set both region-growing probability thresholds to 0.7 (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 62.3/66.8 | Helped |
| use oversegmented superpixels by increasing τ to 0.5 (ablation) | Semantic-Aware Superpixel · PASCAL VOC 2012 val | mIOU: 65.1 -> 62.9 | Hurt |
| use the proposed method rather than AMN | Weakly supervised semantic segmentation · MS-COCO 2014 val | mIOU: 44.7 -> 44.8 | Helped |
| use the proposed method rather than SIPE | Weakly supervised semantic segmentation · PASCAL VOC 2012 test | mIOU: 69.7 -> 70.1 | Helped |
Closest papers by meaning
- Weakly Supervised Semantic Segmentation Using Superpixel Pooling Network (AAAI 2017 · similarity 0.93)
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.90)
- Combining Bottom-Up, Top-Down, and Smoothness Cues for Weakly Supervised Image Segmentation (CVPR 2017 · similarity 0.89)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.88)
- PSDPM: Prototype-based Secondary Discriminative Pixels Mining for Weakly Supervised Semantic Segmentation (CVPR 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.