Research map / Detection and segmentation / CNN features
Deeply Shape-Guided Cascade for Instance Segmentation
Check 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 initial, explicit, and implicit shape guidance to the cascade | HTC · COCO 2017 val, ResNet-50 FPN | box AP: 43.3 -> 45.8 | Helped |
| Add initial, explicit, and implicit shape guidance to the cascade | HTC · COCO 2017 val, ResNet-50 FPN | mask AP: 38.3 -> 40.2 | Helped |
| Add shape-guided cascade components using the same backbone and training strategy | HTC (multi-scale training) · COCO test-dev, ResNeXt-101-64x4d DCN FPN | box AP: 50.8 -> 51.8 | Helped |
| Add shape-guided cascade components using the same backbone and training strategy | HTC (multi-scale training) · COCO test-dev, ResNeXt-101-64x4d DCN FPN | mask AP: 44.2 -> 45.5 | Helped |
| Reduce intermediate mask feature resolution to 7×7 (ablation) | DSC · COCO val | box AP: 44.8 -> 44.5 | Hurt |
| Reduce intermediate mask feature resolution to 7×7 (ablation) | DSC · COCO val | mask AP: 39.5 -> 39.4 | Hurt |
| Add explicit shape guidance through shape-guided RoIAlign (ablation) | F-DSC · COCO val | box AP: 44.2 -> 44.5 | Helped |
| Add explicit shape guidance through shape-guided RoIAlign (ablation) | F-DSC · COCO val | mask AP: 39.1 -> 39.5 | Helped |
| Add implicit shape guidance through feature fusion (ablation) | F-DSC · COCO val | box AP: 43.4 -> 44.5 | Helped |
| Add implicit shape guidance through feature fusion (ablation) | F-DSC · COCO val | mask AP: 38.8 -> 39.5 | Helped |
| Retain the +1 term in shape-guided RoIAlign (ablation) | F-DSC · COCO val | box AP: 44.3 -> 44.5 | Helped |
| Retain the +1 term in shape-guided RoIAlign (ablation) | F-DSC · COCO val | mask AP: 39.3 -> 39.5 | Helped |
Closest papers by meaning
- Hybrid Task Cascade for Instance Segmentation (CVPR 2019 · similarity 0.89)
- Mask Encoding for Single Shot Instance Segmentation (CVPR 2020 · similarity 0.88)
- Instance-Aware Semantic Segmentation via Multi-Task Network Cascades (CVPR 2016 · similarity 0.88)
- Any3DIS: Class-Agnostic 3D Instance Segmentation by 2D Mask Tracking (CVPR 2025 · similarity 0.86)
- MaskClustering: View Consensus based Mask Graph Clustering for Open-Vocabulary 3D Instance Segmentation (CVPR 2024 · similarity 0.85)
- Sketchy Bounding-box Supervision for 3D Instance Segmentation (CVPR 2025 · 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.