Research map / Detection and segmentation / CNN features
Occlusion-Shared and Feature-Separated Network for Occlusion Relationship Reasoning
Rui Lu, Feng Xue, Menghan Zhou, Anlong Ming, Yu Zhou
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 |
|---|---|---|---|
| OFNet vs DOOBNet | OFNet · PIOD OPR | AP: .683 -> .729 | Improved |
| OFNet vs DOOBNet | OFNet · BSDS ownership OPR | AP: 2.8% ODS, 3.7% OIS and 6.1% AP | Improved |
| OFNet vs DOOBNet | OFNet · PIOD EPR | AP: .723 -> .773 | Improved |
| split the decoder into two paths without separate high-level features (ablation) | DOOBNet · PIOD EPR | AP: .723 -> .694 | Hurt |
| use separate high-level features for edge and orientation paths instead of a single edge stream (ablation) | OFNet · PIOD EPR | AP: .685 -> .773 | Improved |
| train orientation path jointly with occlusion cue instead of alone (ablation) | OFNet · PIOD OPR | AP: .674 -> .729 | Improved |
| add low-level cues to the edge path (ablation) | OFNet · PIOD EPR | AP: .764 -> .773 | Improved |
| add high-level cues to the edge path (ablation) | OFNet · PIOD EPR | AP: .758 -> .773 | Improved |
| add high-level (bilateral) cues to the orientation path (ablation) | OFNet · PIOD OPR | AP: .703 -> .729 | Improved |
| add high-level cues to the edge path (ablation) | OFNet · PIOD OPR | AP: .698 -> .729 | Improved |
| increase bilateral:occlusion channel ratio from 16:16 to 64:16 (ablation) | BRF module · PIOD EPR | AP: .749 -> .773 | Improved |
| increase bilateral share further from 64:16 to 80:16 (ablation) | BRF module · PIOD EPR | AP: .773 -> .764 | Hurt |
Closest papers by meaning
- MT-ORL: Multi-Task Occlusion Relationship Learning (ICCV 2021 · similarity 0.92)
- DOC: Deep OCclusion Estimation from a Single Image (ECCV 2016 · similarity 0.89)
- Peek-a-Boo: Occlusion Reasoning in Indoor Scenes With Plane Representations (CVPR 2020 · similarity 0.88)
- Robust Instance Segmentation Through Reasoning About Multi-Object Occlusion (CVPR 2021 · similarity 0.88)
- Pixel-Pair Occlusion Relationship Map (P2ORM): Formulation, Inference & Application (ECCV 2020 · similarity 0.87)
- Amodal Scene Analysis via Holistic Occlusion Relation Inference and Generative Mask Completion (AAAI 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.
