Research map / Video / Video segmentation and tracking
Learning Motion Patterns in Videos
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 GT angle field instead of GT flow vectors (ablation) | MP-Net · DAVIS | Mean IoU: 44.3 -> 46.6 | Helped |
| Use GT angle field instead of GT flow vectors (ablation) | MP-Net · FT3D test | Mean IoU: 74.5 -> 73.1 | Hurt |
| Add RGB input to the GT angle field (ablation) | MP-Net · DAVIS | Mean IoU: 46.6 -> 39.6 | Hurt |
| Train using estimated LDOF flow instead of ground-truth flow (ablation) | MP-Net · FT3D test | Mean IoU: 74.5 -> 63.2 | Hurt |
| Train using estimated LDOF flow instead of ground-truth flow (ablation) | MP-Net · DAVIS | Mean IoU: 46.6 -> 38.1 | Hurt |
| Use four decoder units instead of one (ablation) | MP-Net · FT3D test | Mean IoU: +12.8% | Helped |
| Use four decoder units instead of one (ablation) | MP-Net · DAVIS | Mean IoU: +5.8% | Helped |
| Add objectness cues (ablation) | MP-Net with LDOF flow · DAVIS | Mean IoU: 52.4 -> 63.3 | Helped |
| Add objectness cues (ablation) | MP-Net with EpicFlow · DAVIS | Mean IoU: 56.9 -> 64.5 | Helped |
| Add CRF refinement to MP-Net with objectness (ablation) | MP-Net with LDOF flow · DAVIS | Mean IoU: 63.3 -> 69.7 | Helped |
| Add CRF refinement to MP-Net with objectness (ablation) | MP-Net with EpicFlow · DAVIS | Mean IoU: 64.5 -> 68.0 | Helped |
| Use EpicFlow instead of LDOF as the input flow (ablation) | MP-Net · DAVIS | Mean IoU: 52.4 -> 56.9 | Helped |
Closest papers by meaning
- Learning Segmentation from Point Trajectories (NeurIPS 2024 · similarity 0.91)
- Unsupervised Multi-Object Segmentation by Predicting Probable Motion Patterns (NeurIPS 2022 · similarity 0.91)
- Discovering Objects That Can Move (CVPR 2022 · similarity 0.90)
- Segment Any Motion in Videos (CVPR 2025 · similarity 0.88)
- FlowFeat: Pixel-Dense Embedding of Motion Profiles (NeurIPS 2025 · similarity 0.88)
- GMOS: Grounding Moving Object Segmentation in 3D Space and Time (NeurIPS 2026 · 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.