Research map / Detection and segmentation / Weakly supervised detection and segmentation
When Specialists Meet Generalists: Segmenter-Coordinated Asymmetric Learning for Label-Deficient Concealed Object Segmentation
Chunming He, Dingming Zhang, Longxiang Tang, Ziyun Yang, Fengyang Xiao, Sina Farsiu
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 Phase II (SAM updated by segmenter feedback) (ablation) | SCALER · COD10K (scribble) | F-beta: 0.723 -> 0.737 | Improved |
| add entropy-based image-level weighting E (ablation) | SCALER · COD10K (scribble) | F-beta: 0.724 -> 0.737 | Improved |
| add uncertainty-based pixel-level weighting U (ablation) | SCALER · COD10K (scribble) | F-beta: 0.721 -> 0.737 | Improved |
| add fused SAM pseudo-label PL_G (ablation) | SCALER · COD10K (scribble) | F-beta: 0.732 -> 0.737 | Improved |
| add augmentation invariance loss L_ai (ablation) | SCALER · COD10K (scribble) | F-beta: 0.733 -> 0.737 | Improved |
| add noise resistance loss L_nr (ablation) | SCALER · COD10K (scribble) | F-beta: 0.729 -> 0.737 | Improved |
| pair SAM with a segmenter instead of a second SAM (heterogeneous vs SAM+SAM co-training) (ablation) | SCALER · COD10K (scribble, ResNet50 + SAM) | F-beta: 0.720 -> 0.765 | Improved |
| replace homogeneous Seg+Seg with heterogeneous Seg+SAM (ablation) | CPS co-training · COD10K (scribble) | F-beta: 0.704 -> 0.716 | Improved |
| use SCALER asymmetric optimization on heterogeneous pair (ablation) | CPS co-training · COD10K (scribble) | F-beta: 0.716 -> 0.728 | Improved |
| add Phase II on top of WS-SAM (ablation) | WS-SAM · COD10K (scribble, PVT V2 + SAM3) | F-beta: 0.751 -> 0.775 | Improved |
| use weak-weak augmentation instead of strong-weak (ablation) | SCALER Phase II · COD10K (scribble) | F-beta: 0.727 -> 0.737 | Improved |
| add basic loss R_b (ablation) | SCALER · COD10K (scribble) | F-beta: 0.725 -> 0.737 | Improved |
Closest papers by meaning
- Segment Concealed Objects With Incomplete Supervision (TPAMI 2025 · similarity 0.91)
- Weakly-Supervised Concealed Object Segmentation with SAM-based Pseudo Labeling and Multi-scale Feature Grouping (NeurIPS 2023 · similarity 0.90)
- UniT: Unified Knowledge Transfer for Any-Shot Object Detection and Segmentation (CVPR 2021 · similarity 0.87)
- RUN: Reversible Unfolding Network for Concealed Object Segmentation (ICML 2025 · similarity 0.87)
- Segment Anything without Supervision (NeurIPS 2024 · similarity 0.86)
- Bridge the Points: Graph-based Few-shot Segment Anything Semantically (NeurIPS 2024 · similarity 0.85)
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.
