Research map / Detection and segmentation / Open-set and few-shot detection
E-SAM: Training-Free Segment Every Entity Model
Weiming Zhang, Dingwen Xiao, Lei Chen, Lin Wang
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 MMG to AMG baseline (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 17.3 -> 20.3 | Improved |
| add MMG on top of EMR and USR (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 40.8 -> 43.6 | Improved |
| add EMR to AMG baseline (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 17.3 -> 29.5 | Improved |
| add EMR on top of MMG and USR (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 38.1 -> 43.6 | Improved |
| add USR to AMG baseline (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 17.3 -> 22.7 | Improved |
| add USR on top of MMG and EMR (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 35.0 -> 43.6 | Improved |
| add USR to MMG only (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP_L^e: 20.3 -> 38.1 | Improved |
| use SAM object-level masks only instead of all-level AMG | SAM · EntitySeg-LR | AP_L^e: 17.0 -> 23.7 | Improved |
| increase point prompts per side from 16/16 to 32/64 (ablation) | E-SAM · EntitySeg-LR (ViT-L) | AP^e: 39.8 -> 43.6 | Improved |
| tune MMG threshold theta_O to 0.8 (ablation) | MMG · EntitySeg-LR (ViT-L) | AP^e: 19.5 -> 20.3 | Mixed |
| add training-free E-SAM (ViT-H) vs SAM AMG (ViT-H) | SAM · EntitySeg val | AP^e: 20.1 -> 50.2 | Improved |
| E-SAM ViT-L vs CropFormer Swin-L | E-SAM · EntitySeg val | AP^e: 48.0 -> 45.9 | Hurt |
Closest papers by meaning
- EntitySAM: Segment Everything in Video (CVPR 2025 · similarity 0.88)
- SOHES: Self-supervised Open-world Hierarchical Entity Segmentation (ICLR 2024 · similarity 0.88)
- Segment Anything without Supervision (NeurIPS 2024 · similarity 0.87)
- Seg2Any: Open-set Segmentation-Mask-to-Image Generation with Precise Shape and Semantic Control (NeurIPS 2025 · similarity 0.87)
- Personalize Segment Anything Model with One Shot (ICLR 2024 · similarity 0.87)
- Effective SAM Combination for Open-Vocabulary Semantic Segmentation (CVPR 2025 · 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.
