Research map / 3D vision / Depth and stereo
Pixel-Pair Occlusion Relationship Map (P2ORM): Formulation, Inference & Application
Xuchong Qiu, Yang Xiao, Chaohui Wang, Renaud Marlet
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 pixel-pair occlusion relationship (4-connectivity) instead of separate boundary and orientation heads | ResNet U-Net · BSDS ownership | ODS: .571 -> .590 | Improved |
| use pixel-pair occlusion relationship (4-connectivity) instead of separate boundary and orientation heads | ResNet U-Net · NYUv2-OR | ODS: .396 -> .500 | Improved |
| use pixel-pair occlusion relationship (4-connectivity) instead of separate boundary and orientation heads | ResNet U-Net · iBims-1-OR | ODS: .482 -> .575 | Improved |
| add diagonal neighbors (8-connectivity instead of 4) (ablation) | P2ORM · BSDS ownership | ODS: .590 -> .607 | Improved |
| add diagonal neighbors (8-connectivity instead of 4) (ablation) | P2ORM · NYUv2-OR | ODS: .500 -> .520 | Improved |
| add diagonal neighbors (8-connectivity instead of 4) (ablation) | P2ORM · iBims-1-OR | ODS: .575 -> .581 | Improved |
| add diagonal neighbors (8-connectivity instead of 4) (ablation) | P2ORM · NYUv2-OR | AP: .477 -> .497 | Improved |
| use plain ResNet U-Net baseline instead of specialized occlusion architectures | baseline variant · BSDS ownership | ODS: .553 -> .571 | No clear effect |
| refine SharpNet depth with estimated P2ORM | depth refinement · NYUv2 and iBims-1 | depth-edge accuracy and completion | Improved |
| refine depth with P2ORM | depth refinement · NYUv2 and iBims-1 | general depth metrics (rel, RMSE): <1% | No clear effect |
| use the proposed refinement variant over alternatives (extra inputs, GT-depth loss, d_pq only, D_pq only) (ablation) | depth refinement · iBims-1 / NYUv2 | depth-edge metrics | Improved |
Closest papers by meaning
- Predicting Sharp and Accurate Occlusion Boundaries in Monocular Depth Estimation Using Displacement Fields (CVPR 2020 · similarity 0.88)
- Occlusion-Aware Depth Estimation with Adaptive Normal Constraints (ECCV 2020 · similarity 0.88)
- DOC: Deep OCclusion Estimation from a Single Image (ECCV 2016 · similarity 0.87)
- MT-ORL: Multi-Task Occlusion Relationship Learning (ICCV 2021 · similarity 0.87)
- Occlusion-Shared and Feature-Separated Network for Occlusion Relationship Reasoning (ICCV 2019 · similarity 0.87)
- Amodal Scene Analysis via Holistic Occlusion Relation Inference and Generative Mask Completion (AAAI 2024 · 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.
