Research map / Detection and segmentation / Open-set and few-shot detection
Robust Motion Segmentation From Pairwise Matches
Federica Arrigoni, 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 |
|---|---|---|---|
| replace spanning-tree baseline with mode voting over redundant pairwise segmentations | pairwise segmentation with RPA · Hopkins155 (all sequences) | misclassification error (%): 3.79 -> 1.37 | Improved |
| replace spanning-tree baseline with mode voting over redundant pairwise segmentations | pairwise segmentation with RPA · Hopkins12 | mean misclassification error (%): 7.45 -> 4.33 | Improved |
| replace spanning-tree baseline with mode voting | pairwise segmentation with RPA · Hopkins12 | median misclassification error (%): 2.16 -> 0.38 | Improved |
| use Mode instead of the baseline on indoor sequences | pairwise segmentation with RPA · Indoor sequences | misclassification error (%): 0.76 vs 0.95; 3.80 vs 2.30 | No clear effect |
| use Mode instead of the baseline on indoor sequences | pairwise segmentation with RPA · Indoor sequence Penguin | classified points (%): 33.95 -> 69.17 | Improved |
| use Mode instead of the baseline on helicopter | pairwise segmentation with RPA · Outdoor sequence helicopter | classified points (%): 45.93 -> 80.82 | Improved |
| use Mode instead of the baseline on helicopter | pairwise segmentation with RPA · Outdoor sequence helicopter | misclassification error (%): 0.78 -> 2.01 | Hurt |
| use Mode instead of track-based Subset on helicopter | pairwise matches input · Outdoor sequence helicopter | misclassification error (%): 16.81 -> 2.01 | Improved |
| use Mode instead of track-based Subset on noisy indoor Penguin | pairwise matches input · Indoor sequence Penguin | misclassification error (%): 32.27 -> 0.76 | Improved |
| increase fraction of mismatched correspondences up to 60% | Mode · cars1 simulated | misclassification error (%): around 0% up to 60% mismatches | No clear effect |
| increase mismatches above 10% for track-based Subset and RSIM | Subset and RSIM · cars1 simulated | misclassification error (%) | Hurt |
| compute tracks with QuichMatch instead of StableSfM (ablation) | Subset and RSIM · indoor sequences | misclassification error (%): 99.59 -> 70.11 classified | No clear effect |
Closest papers by meaning
- Multi-frame Motion Segmentation by Combining Two-Frame Results (IJCV 2022 · similarity 0.87)
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.87)
- What to Select: Pursuing Consistent Motion Segmentation from Multiple Geometric Models (AAAI 2021 · similarity 0.87)
- Unsupervised Multi-Object Segmentation by Predicting Probable Motion Patterns (NeurIPS 2022 · similarity 0.86)
- The Best of Both Worlds: Combining CNNs and Geometric Constraints for Hierarchical Motion Segmentation (CVPR 2018 · similarity 0.86)
- Unsupervised Space-Time Network for Temporally-Consistent Segmentation of Multiple Motions (CVPR 2023 · 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.
