Research map / Detection and segmentation / CNN features
Occlusion Boundary Detection via Deep Exploration of Context
Huan Fu, Chaohui Wang, Dacheng Tao, Michael J. Black
Check 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 the proposed CNN-CRF detector with probabilistic output | Occlusion boundary detection · CMU benchmark | F-measure: 0.62 -> 0.71 | Improved |
| Add both motion features OMF1 and OMF2 to image cues (ablation) | Occlusion boundary detector · CMU benchmark | F-measure: 0.60 -> 0.71 | Improved |
| Add OMF2 while retaining OMF1 (ablation) | Occlusion boundary detector · CMU benchmark | F-measure: 0.65 -> 0.71 | Improved |
| Add OMF1 while retaining OMF2 (ablation) | Occlusion boundary detector · CMU benchmark | F-measure: 0.64 -> 0.71 | Improved |
| Replace Classic-NL with LDOF optical flow (ablation) | Occlusion boundary detector · CMU benchmark | F-measure: 0.71 -> 0.68 | Hurt |
| Replace Classic-NL with Deepflow optical flow (ablation) | Occlusion boundary detector · CMU benchmark | F-measure: 0.71 -> 0.69 | Hurt |
| Replace Classic-NL with Sun et al. optical flow (ablation) | Occlusion boundary detector · CMU benchmark | F-measure: 0.71 | No clear effect |
| Map a large patch to its center pixel instead of a small structured patch (ablation) | CNN context model · CMU benchmark | F-measure: 0.71 -> 0.66 | Hurt |
| Predict each pixel in the small output patch independently (ablation) | CNN context model · CMU benchmark | F-measure: 0.71 -> 0.65 | Hurt |
| Predict an output patch the same size as the input patch (ablation) | CNN context model · CMU benchmark | F-measure: 0.71 -> 0.66 | Hurt |
| Use a small patch as both the CNN input and output (ablation) | CNN context model · CMU benchmark | F-measure: 0.71 -> 0.63 | Hurt |
Closest papers by meaning
- Occlusion Boundary: A Formal Definition & Its Detection via Deep Exploration of Context (TPAMI 2020 · similarity 0.96)
- DOC: Deep OCclusion Estimation from a Single Image (ECCV 2016 · similarity 0.89)
- Local Detection of Stereo Occlusion Boundaries (CVPR 2019 · similarity 0.89)
- Occluded Video Instance Segmentation: A Benchmark (IJCV 2022 · similarity 0.87)
- Learning on the Edge: Investigating Boundary Filters in CNNs (IJCV 2019 · similarity 0.87)
- Convolutional Oriented Boundaries: From Image Segmentation to High-Level Tasks (TPAMI 2017 · 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.
