Research map / 3D vision / Depth and stereo
Motion Segmentation by Exploiting Complementary Geometric Models
Xun Xu, Loong Fah Cheong, Zhuwen Li
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 |
|---|---|---|---|
| Fuse the affine, homography, and fundamental-matrix views with subset constraints instead of using one geometric model | Motion segmentation · Hopkins155 | classification error (%): 0.59 -> 0.31 | Improved |
| Fuse the affine, homography, and fundamental-matrix views with kernel addition instead of using one geometric model | Motion segmentation · Hopkins155 | classification error (%): 0.59 -> 0.36 | Improved |
| Use subset-constrained fusion rather than the single affine view | Motion segmentation · Hopkins12 | classification error (%): 0.15 -> 0.06 | Improved |
| Use subset-constrained fusion rather than the single affine view | Motion segmentation · MTPV62 perspective (9 clips) | classification error (%): 0.25 -> 0.20 | Improved |
| Use multi-view co-regularization rather than the single homography view | Motion segmentation · KT3DMoSeg | average classification error (%): 11.45 -> 7.92 | Improved |
| Use multi-view co-regularization rather than the single homography view | Motion segmentation · KT3DMoSeg | median classification error (%): 7.14 -> 0.75 | Improved |
| Replace subset-constrained fusion with co-regularization (ablation) | Multi-view spectral clustering · Hopkins155 | classification error (%): 0.31 -> 0.46 | Hurt |
| Replace subset-constrained fusion with kernel addition (ablation) | Multi-view spectral clustering · Hopkins155 | classification error (%): 0.31 -> 0.36 | Hurt |
| Use the fundamental-matrix view alone instead of fused views (ablation) | Motion segmentation · Hopkins155 | classification error (%): 0.31 -> 1.79 | Hurt |
| Use the fundamental-matrix view alone instead of fused views (ablation) | Motion segmentation · MTPV62 perspective (9 clips) | classification error (%): 0.20 -> 5.09 | Hurt |
| Vary co-regularization strength to modify the fundamental-matrix view (ablation) | Co-regularized F-view · KT3DMoSeg | classification error (%) | Mixed |
Closest papers by meaning
- 3D Rigid Motion Segmentation with Mixed and Unknown Number of Models (TPAMI 2019 · similarity 0.96)
- What to Select: Pursuing Consistent Motion Segmentation from Multiple Geometric Models (AAAI 2021 · similarity 0.91)
- Graph Constrained Data Representation Learning for Human Motion Segmentation (ICCV 2021 · similarity 0.87)
- Multi-frame Motion Segmentation by Combining Two-Frame Results (IJCV 2022 · similarity 0.87)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.87)
- On the Usage of the Trifocal Tensor in Motion Segmentation (ECCV 2020 · 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.
