Research map / Detection and segmentation / Open-set and few-shot detection
Amodal Instance Segmentation
Ke Li, Jitendra Malik
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 |
|---|---|---|---|
| train on synthetic occlusion for amodal masks instead of modal segmentation (IIS) | Faster R-CNN pipeline · 100 annotated VOC objects | mAP^r at 50% IoU: 34.1 -> 45.2 | Improved |
| train on synthetic occlusion for amodal masks instead of modal segmentation (IIS) | Faster R-CNN pipeline · 100 annotated VOC objects | mAP^r at 70% IoU: 14.0 -> 22.6 | Improved |
| amodal model instead of IIS with GT boxes | segmentation · 100 annotated VOC objects | accuracy at IoU 50%: 68.0 -> 80.0 | Improved |
| amodal model instead of IIS with GT boxes | segmentation · 100 annotated VOC objects | area under accuracy curve: 57.5 -> 64.3 | Improved |
| amodal model instead of IIS | Faster R-CNN pipeline · PASCAL 3D+ rigid objects | mAP^r at 50% IoU: 37.4 -> 44.0 | Improved |
| amodal model instead of IIS | Faster R-CNN pipeline · PASCAL 3D+ rigid objects | mAP^r at 70% IoU: 15.9 -> 20.9 | Improved |
| add modal segmentation heatmap as input (ablation) | proposed method · 100 annotated VOC objects | mAP^r at 50% IoU: 35.2 -> 45.2 | Improved |
| add modal segmentation heatmap as input (ablation) | proposed method · 100 annotated VOC objects | mAP^r at 70% IoU: 18.4 -> 22.6 | Improved |
| add dynamic sample generation of varied occlusions (ablation) | proposed method · 100 annotated VOC objects | mAP^r at 50% IoU: 39.8 -> 45.2 | Improved |
| add dynamic sample generation of varied occlusions (ablation) | proposed method · 100 annotated VOC objects | mAP^r at 70% IoU: 22.7 -> 22.6 | No clear effect |
Closest papers by meaning
- Amodal Instance Segmentation via Prior-Guided Expansion (AAAI 2023 · similarity 0.91)
- Amodal Ground Truth and Completion in the Wild (CVPR 2024 · similarity 0.89)
- Using Diffusion Priors for Video Amodal Segmentation (CVPR 2025 · similarity 0.89)
- Towards Efficient Foundation Model for Zero-shot Amodal Segmentation (CVPR 2025 · similarity 0.89)
- Segment Anything, Even Occluded (CVPR 2025 · similarity 0.88)
- BLADE: Box-Level Supervised Amodal Segmentation through Directed Expansion (AAAI 2024 · similarity 0.88)
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.
