Research map / Detection and segmentation / Weakly supervised detection and segmentation
Modeling the Label Distributions for Weakly-Supervised Semantic Segmentation
Linshan Wu, Zhun Zhong, Jiayi Ma, Yunchao Wei, Hao Chen, Leyuan Fang, Shutao Li A
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 L_self consistency between segmentation and GMM predictions (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 63.2 -> 72.0 | Improved |
| add L_weak (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 72.0 -> 74.5 | Improved |
| add L_con (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 74.5 -> 75.0 | Improved |
| add OEM optimization (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 75.0 -> 75.6 | Improved |
| add L_self (ablation) | AGMM++ · PASCAL scribble-supervised | mIoU: 69.0 -> 73.2 | Improved |
| add OEM optimization (ablation) | AGMM++ · PASCAL box-supervised | mIoU: 74.5 -> 75.5 | Improved |
| add OEM and redefined L_con over AGMM (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 69.6 -> 72.3 | Improved |
| add OEM and redefined L_con over AGMM (ablation) | AGMM++ · PASCAL scribble-supervised | mIoU: 76.4 -> 77.5 | Improved |
| add L_self to pseudo-label generation over ToCo CAMs (ablation) | AGMM++ · PASCAL image-level pseudo labels | pseudo label mIoU: 70.5 -> 71.5 | Improved |
| add OEM to pseudo label generation (ablation) | AGMM++ · PASCAL image-level pseudo labels | pseudo label mIoU: 72.5 -> 73.3 | Improved |
| add stop-gradient to GMM (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 72.3 -> 67.4 | Hurt |
| use hard one-hot online labels instead of soft GMM labels (ablation) | AGMM++ · PASCAL point-supervised | mIoU: 72.3 -> 66.5 | Hurt |
Closest papers by meaning
- Employing Multi-Estimations for Weakly-Supervised Semantic Segmentation (ECCV 2020 · similarity 0.91)
- W2P: Switching from Weak Supervision to Partial Supervision for Semantic Segmentation (AAAI 2024 · similarity 0.91)
- Learning Affinity From Attention: End-to-End Weakly-Supervised Semantic Segmentation With Transformers (CVPR 2022 · similarity 0.90)
- Re-Distributing Biased Pseudo Labels for Semi-Supervised Semantic Segmentation: A Baseline Investigation (ICCV 2021 · similarity 0.90)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.90)
- WISH: Weakly Supervised Instance Segmentation using Heterogeneous Labels (CVPR 2025 · similarity 0.89)
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.
