Research map / Detection and segmentation / Tracking
AnthroTAP: Learning Point Tracking with Real-World Motion
Inès Hyeonsu Kim, Seokju Cho, Jahyeok Koo, Junghyun Park, Jiahui Huang, Honglak Lee, Joon-Young Lee, Seungryong Kim
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 |
|---|---|---|---|
| fine-tune with AnthroTAP human-motion pseudo-labels (Kub+1.4K videos) | LocoTrack · TAP-Vid DAVIS First (256x256) | AJ: 63.0 -> 64.8 | Improved |
| fine-tune with AnthroTAP pseudo-labels | TAPNext · TAP-Vid DAVIS Strided (256x256) | AJ: 65.4 -> 71.4 | Improved |
| fine-tune with AnthroTAP pseudo-labels | TAPNext · RoboTAP First | AJ: 59.8 -> 63.4 (BootsTAPNext 64.1) | Mixed |
| add optical flow based track rejection to pseudo-label pipeline (ablation) | AnthroTAP (LocoTrack) · TAP-Vid DAVIS | AJ: 64.4 -> 64.8 | Improved |
| use AnthroTAP pseudo-labels instead of CoTracker3 self-training on same Let's Dance videos (ablation) | CoTracker3 · TAP-Vid DAVIS | AJ: 64.2 -> 65.0 | Improved |
| train on real pseudo-labeled data instead of synthetic perfect-GT BlinkVision data (ablation) | LocoTrack · TAP-Vid DAVIS | AJ: 64.2 -> 64.8 | Improved |
| training on human-only motion data, evaluated on non-human points (ablation) | Anthro-LocoTrack · TAP-Vid DAVIS non-human points | AJ: 58.8 -> 60.8 | Improved |
| training on human-only motion data, evaluated on human points (ablation) | Anthro-LocoTrack · TAP-Vid DAVIS human points | AJ: 50.7 -> 51.2 | Improved |
Closest papers by meaning
- PoseTrack: A Benchmark for Human Pose Estimation and Tracking (CVPR 2018 · similarity 0.88)
- Motions as Queries: One-Stage Multi-Person Holistic Human Motion Capture (CVPR 2025 · similarity 0.87)
- Is This Tracker On? A Benchmark Protocol for Dynamic Tracking (NeurIPS 2025 · similarity 0.86)
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.86)
- HumanVid: Demystifying Training Data for Camera-controllable Human Image Animation (NeurIPS 2024 · similarity 0.86)
- DriveTrack: A Benchmark for Long-Range Point Tracking in Real-World Videos (CVPR 2024 · 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.
