Research map / Detection and segmentation / Weakly supervised detection and segmentation
Towards Noiseless Object Contours 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 |
|---|---|---|---|
| Use SANCE-generated pseudo labels instead of the prior method's labels | DeepLab-ASPP on PASCAL VOC 2012 · PASCAL VOC 2012 test | mIoU: 70.0 -> 72.2 | Helped |
| Use external saliency maps with SANCE | SANCE on PASCAL VOC 2012 · PASCAL VOC 2012 val | mIoU: 70.9 -> 72.0 | Helped |
| Use SANCE-generated pseudo labels instead of RIB labels | DeepLab-ASPP on MS COCO 2014 · MS COCO 2014 val | mIoU: 43.8 -> 44.7 | Helped |
| Add online-label supervision to the contour branch (ablation) | SANCE-naive on PASCAL VOC 2012 · PASCAL VOC 2012 train | pseudo-label mIoU: 67.8 -> 71.3 | Helped |
| Add online-label supervision to both contour and segmentation branches (ablation) | SANCE-naive on PASCAL VOC 2012 · PASCAL VOC 2012 val | mIoU: 66.3 -> 69.6 | Helped |
| Use segmentation map rather than CAM to generate pseudo labels (ablation) | SANCE on PASCAL VOC 2012 · PASCAL VOC 2012 train | pseudo-label mIoU: 70.7 -> 74.6 | Helped |
| Add the label refine module (ablation) | SANCE on PASCAL VOC 2012 · PASCAL VOC 2012 train | pseudo-label mIoU: 74.6 -> 75.7 | Helped |
| Add multi-scale augmentation for pseudo-label generation (ablation) | SANCE on PASCAL VOC 2012 · PASCAL VOC 2012 val | mIoU: 69.9 -> 70.9 | Helped |
| Use stride 8 rather than stride 16 (ablation) | SANCE-naive on PASCAL VOC 2012 · PASCAL VOC 2012 train | pseudo-label mIoU: 69.9 -> 70.7 | Helped |
| Refine online labels for both branches rather than only the contour branch (ablation) | SANCE on PASCAL VOC 2012 · PASCAL VOC 2012 train | pseudo-label mIoU: 76.9 -> 75.7 | Hurt |
Closest papers by meaning
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.89)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.89)
- Single-Stage Semantic Segmentation From Image Labels (CVPR 2020 · similarity 0.88)
- From SAM to CAMs: Exploring Segment Anything Model for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.88)
- SFC: Shared Feature Calibration in Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.87)
- Frozen CLIP: A Strong Backbone for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.86)
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.