Research map / Detection and segmentation / Open-set and few-shot detection
Amodal Instance Segmentation via Prior-Guided Expansion
Junjie Chen, Li Niu, Jianfu Zhang, Jianlou Si, Chen Qian, Liqing Zhang
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 |
|---|---|---|---|
| Use prior-guided box-level and pixel-level expansion | Mask R-CNN · KINS | amodal mask AP: 32.20* -> 33.82 | Improved |
| Use prior-guided box-level and pixel-level expansion | Mask R-CNN · D2SA | amodal mask AP: 70.27* -> 71.79 | Improved |
| Use prior-guided box-level and pixel-level expansion | Mask R-CNN · COCOA cls | amodal mask AP: 35.41* -> 37.55 | Improved |
| Add the prior-guided amodal box head (ablation) | Mask R-CNN · KINS | amodal box AP: 32.54 -> 35.63 | Improved |
| Add the prior-guided amodal box head (ablation) | Mask R-CNN · KINS | amodal mask AP: 30.12 -> 31.95 | Improved |
| Add the prior-guided amodal mask head (ablation) | Mask R-CNN · KINS | amodal mask AP: 30.12 -> 31.25 | Improved |
| Add both prior-guided amodal box and mask heads (ablation) | Mask R-CNN · KINS | amodal mask AP: 30.12 -> 33.82 | Improved |
| Add both prior-guided amodal box and mask heads (ablation) | Mask R-CNN · KINS | amodal box AP: 32.54 -> 35.77 | Improved |
| Add a box head without expansion prior (ablation) | Mask R-CNN · KINS | amodal mask AP: 30.12 -> 30.66 | Improved |
| Increase the number of retrieved expansion priors (ablation) | PGExp · KINS | amodal box and mask performance | Mixed |
| Increase the memory-bank size (ablation) | PGExp · KINS | amodal box and mask performance | No clear effect |
Closest papers by meaning
- BLADE: Box-Level Supervised Amodal Segmentation through Directed Expansion (AAAI 2024 · similarity 0.93)
- Amodal Instance Segmentation (ECCV 2016 · similarity 0.91)
- Amodal Ground Truth and Completion in the Wild (CVPR 2024 · similarity 0.90)
- Segment Anything, Even Occluded (CVPR 2025 · similarity 0.89)
- Using Diffusion Priors for Video Amodal Segmentation (CVPR 2025 · similarity 0.88)
- Towards Efficient Foundation Model for Zero-shot Amodal Segmentation (CVPR 2025 · similarity 0.87)
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.
