Research map / Detection and segmentation / Weakly supervised detection and segmentation
Box-Driven Class-Wise Region Masking and Filling Rate Guided Loss 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 box-driven class-wise masking and refined filling-rate loss | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val, weak CRF proposals | mIoU: 60.6 -> 66.8 | Helped |
| Add box-driven class-wise masking (ablation) | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val | mIoU: 60.6 -> 65.6 | Helped |
| Use box-driven class-wise masking instead of global masking (ablation) | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val | mIoU: 64.9 -> 65.6 | Helped |
| Add class-specific filling-rate guided loss (ablation) | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val | mIoU: 60.6 -> 65.8 | Helped |
| Replace class-specific filling rates with one global filling rate of 0.6 (ablation) | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val | mIoU: 65.8 -> 64.1 | Hurt |
| Refine filling rates by clustering each class into sub-classes (ablation) | DeepLab-LargeFOV VGG-16 with FR-loss · PASCAL VOC 2012 val | mIoU: 65.8 -> 66.3 | Helped |
| Combine BCM with refined FR-loss (ablation) | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val | mIoU: 65.6; 66.3 -> 66.8 | Helped |
| Add box-driven class-wise masking without box supervision (ablation) | DeepLab-LargeFOV VGG-16 · PASCAL VOC 2012 val | mIoU: 60.6 -> 63.4 | Helped |
| Use rectangle-box supervision rather than CRF proposals | Ours · PASCAL VOC 2012 val, weak supervision | mIoU: 66.8 -> 54.9 | Hurt |
| Add 1,464 ground-truth-labeled samples | Ours_CRF · PASCAL VOC 2012 val, semi-supervised | mIoU: 66.8 -> 67.5 | Helped |
| Use the proposed method instead of SDI | PASCAL VOC 2012 val, weak supervision with ResNet-101 · PASCAL VOC 2012 val | mIoU: 69.4 -> 70.2 | Helped |
| Use the proposed method instead of WSSL with CRF proposals | PASCAL VOC 2012 val, weak supervision · PASCAL VOC 2012 val | mIoU: 60.6 -> 66.8 | Helped |
Closest papers by meaning
- Weak-shot Semantic Segmentation via Dual Similarity Transfer (NeurIPS 2022 · similarity 0.91)
- BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation (CVPR 2023 · similarity 0.90)
- Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach (AAAI 2020 · similarity 0.90)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.90)
- Background-Aware Pooling and Noise-Aware Loss for Weakly-Supervised Semantic Segmentation (CVPR 2021 · similarity 0.90)
- Combining Bottom-Up, Top-Down, and Smoothness Cues for Weakly Supervised Image Segmentation (CVPR 2017 · similarity 0.90)
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.