Research map / Video / Video segmentation and tracking
Tracking Everything Everywhere All at Once
Qianqian Wang, Yen-Yu Chang, Ruojin Cai, Zhengqi Li, Bharath Hariharan, Aleksander Holynski, Noah Snavely
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 invertible bijective mapping instead of separate forward/backward networks (ablation) | OmniMotion · TAP-Vid DAVIS | AJ: 12.5 -> 51.7 | Improved |
| add photometric loss (ablation) | OmniMotion · TAP-Vid DAVIS | AJ: 42.3 -> 51.7 | Improved |
| add hard-example mining sampling instead of uniform sampling (ablation) | OmniMotion · TAP-Vid DAVIS | AJ: 47.8 -> 51.7 | Improved |
| add flow loss (ablation) | OmniMotion · TAP-Vid DAVIS | AJ: 23.9 -> 51.7 | Improved |
| use quasi-3D representation with bijections instead of plain 2D MLP (ablation) | OmniMotion · TAP-Vid DAVIS | AJ: 11.6 -> 51.7 | Improved |
| increase samples per ray from 8 to 32 (ablation) | OmniMotion · TAP-Vid DAVIS | AJ: 48.1 -> 51.7 | Improved |
| samples per ray 16 instead of 32 (ablation) | OmniMotion · TAP-Vid DAVIS | OA: 85.6 vs 85.3 | No clear effect |
| optimize OmniMotion on RAFT pairwise flow instead of using RAFT-D directly | RAFT · TAP-Vid DAVIS | temporal coherence (TC): 9.83 -> 0.74 | Improved |
| optimize OmniMotion on TAP-Net correspondences | TAP-Net · TAP-Vid DAVIS | AJ: 38.4 -> 50.9 | Improved |
| RAFT input vs TAP-Net input to OmniMotion | OmniMotion · TAP-Vid RGB-Stacking | <delta_avg: RAFT-D 85.1 vs Ours (TAP-Net) 84.1 | Mixed |
Closest papers by meaning
- Dense Optical Tracking: Connecting the Dots (CVPR 2024 · similarity 0.89)
- FlowTrack: Revisiting Optical Flow for Long-Range Dense Tracking (CVPR 2024 · similarity 0.88)
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.87)
- DELTA: DENSE EFFICIENT LONG-RANGE 3D TRACKING FOR ANY VIDEO (ICLR 2025 · similarity 0.87)
- SpatialTracker: Tracking Any 2D Pixels in 3D Space (CVPR 2024 · similarity 0.87)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (CVPR 2025 · similarity 0.86)
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.
