Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly-Supervised Concealed Object Segmentation with SAM-based Pseudo Labeling and Multi-scale Feature Grouping
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 |
|---|---|---|---|
| add multi-augmentation result fusion to single-mask SAM pseudo labels (ablation) | WS-SAM · COD10K scribble | S-alpha: 0.737 -> 0.772 | Helped |
| add entropy-based pixel-level weighting (ablation) | WS-SAM · COD10K scribble | S-alpha: 0.772 -> 0.793 | Helped |
| add entropy-based image-level selection (ablation) | WS-SAM · COD10K scribble | S-alpha: 0.793 -> 0.803 | Helped |
| add MFG module (ablation) | segmenter · COD10K scribble | S-alpha: 0.780 -> 0.803 | Helped |
| use feature grouping instead of slot attention (ablation) | MFG · COD10K scribble | S-alpha: 0.797 -> 0.803 | Helped |
| add multi-scale aggregation (ablation) | MFG · COD10K scribble | S-alpha: 0.783 -> 0.803 | Helped |
| use weighted gate mechanism instead of fixed coefficient (ablation) | MFG · COD10K scribble | S-alpha: 0.792 -> 0.803 | Helped |
| integrate SCOD with WS-SAM pseudo labels (SCOD+) | SCOD · COD10K scribble | S-alpha: 0.733 -> 0.742 | Helped |
| use SAM pseudo labels to train segmenter instead of fine-tuning SAM directly | WS-SAM · COD10K scribble | S-alpha: 0.772 -> 0.803 | Helped |
| increase augmented views K from 1 to 12 (ablation) | WS-SAM · COD10K scribble | S-alpha: 0.737 -> 0.803 | Helped |
| increase K from 12 to 18 (ablation) | WS-SAM · COD10K scribble | S-alpha: 0.803 -> 0.800 | No clear effect |
| add MFG to fully supervised baseline | COD baseline · COD10K fully supervised | E-phi: 0.879 -> 0.897 | Helped |
Closest papers by meaning
- Universal Weakly Supervised Segmentation by Pixel-to-Segment Contrastive Learning (ICLR 2021 · similarity 0.88)
- RUN: Reversible Unfolding Network for Concealed Object Segmentation (ICML 2025 · similarity 0.86)
- Weakly Supervised Semantic Segmentation for Driving Scenes (AAAI 2024 · similarity 0.86)
- Segment Anything without Supervision (NeurIPS 2024 · similarity 0.86)
- Progressive Feature Self-Reinforcement for Weakly Supervised Semantic Segmentation (AAAI 2024 · similarity 0.85)
- From SAM to CAMs: Exploring Segment Anything Model for Weakly Supervised Semantic Segmentation (CVPR 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.