Research map / Video / Video segmentation and tracking
On the Usage of the Trifocal Tensor in Motion Segmentation
Federica Arrigoni, Luca Magri, Tomas Pajdla
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 trifocal tensor (TriSeg) instead of fundamental matrix (Mode) for partial segmentation | RPA-based motion segmentation with synchronization · Existing benchmark, 12 sequences | misclassification error (%): lowest error in 9 out of 12 sequences | Improved |
| use trifocal tensor (TriSeg) instead of fundamental matrix (Mode) | RPA-based motion segmentation · Novel benchmark stuffed animals1 | misclassification error (%): 8.92 -> 1.82 | Improved |
| use trifocal tensor (TriSeg) instead of fundamental matrix (Mode) | RPA-based motion segmentation · Novel benchmark stuffed animals2 | misclassification error (%): 17.56 -> 3.05 | Improved |
| add fundamental-matrix pair segmentations to triplets (TriPairSeg vs TriSeg) (ablation) | TriSeg · Existing benchmark (all correspondences) | classified points (%): 60.51 -> 79.56 (Pen) | Improved |
| add fundamental-matrix pair segmentations to triplets (TriPairSeg vs TriSeg) (ablation) | TriSeg · Novel benchmark stuffed animals1 | misclassification error (%): 1.82 -> 4.73 | Hurt |
| add fundamental-matrix pair segmentations to triplets (TriPairSeg vs TriSeg) (ablation) | TriSeg · Existing benchmark, trajectories of length two removed | misclassification error (%): Pouch 1.07 -> 0.65; Tea 0.01 -> 0.09 | Mixed |
| remove trajectories of length two from input correspondences (ablation) | TriSeg · Existing benchmark, Pen | classified points (%): 60.51 -> 83.98 | Improved |
| remove trajectories of length two from input correspondences (ablation) | TriSeg · Existing benchmark, Pen | misclassification error (%): 0.15 -> 0.15 | No clear effect |
| remove trajectories of length two from input correspondences (ablation) | Mode · Existing benchmark, Tea | misclassification error (%): 0.29 -> 0.15 | Improved |
| remove trajectories of length two from input correspondences (ablation) | Synch · Existing benchmark, Pen | misclassification error (%): 0.82 -> 0.42 | Improved |
Closest papers by meaning
- What to Select: Pursuing Consistent Motion Segmentation from Multiple Geometric Models (AAAI 2021 · similarity 0.88)
- Motion Segmentation by Exploiting Complementary Geometric Models (CVPR 2018 · similarity 0.87)
- Tensor-Based Synchronization and the Low-Rankness of the Block Trifocal Tensor (NeurIPS 2024 · similarity 0.86)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.83)
- Layered Motion Fusion: Lifting Motion Segmentation to 3D in Egocentric Videos (CVPR 2025 · similarity 0.83)
- M^3-VOS: Multi-Phase, Multi-Transition, and Multi-Scenery Video Object Segmentation (CVPR 2025 · similarity 0.83)
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.
