Research map / Detection and segmentation / Tracking
Tracklet Association by Online Target-Specific Metric Learning and Coherent Dynamics Estimation
Bing Wang, Gang Wang, Kap Luk Chan, Li Wang
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 common class metric with target-specific metric learning (TSML) (ablation) | tracklet association · PETS2009-S2L1 | MOTA: 92.1% -> 93.4% | Improved |
| replace common class metric with target-specific metric learning (TSML) (ablation) | tracklet association · TUD Crossing | IDS / correctly matched: 10/845 -> 7/862 | Improved |
| replace common class metric with target-specific metric learning (TSML) (ablation) | tracklet association · ETH | ID switches: 19 -> 5 | Improved |
| add tracklet dynamics (TD) to TSML (ablation) | tracklet affinity model · PETS2009-S2L1 | MOTA: 93.4% -> 94.7% | Improved |
| add tracklet dynamics (TD) to TSML (ablation) | tracklet affinity model · Town Centre | MOTA: 57.3% -> 61.6% | Improved |
| add tracklet dynamics (TD) to TSML (ablation) | tracklet affinity model · TUD-Stadtmitte | MOTA: 97.3% -> 97.3% | No clear effect |
| add learned weighting parameters (WP) (ablation) | TSML+TD · PETS2009-S2L1 | MOTA: 94.7% -> 95.3% | Improved |
| add learned weighting parameters (WP) (ablation) | TSML+TD · Town Centre | MOTA: 61.6% -> 66.8% | Improved |
| add learned weighting parameters (WP) (ablation) | TSML+TD · TUD-Stadtmitte | MOTA: 97.3% -> 97.3% | No clear effect |
| add one round of tracklet refinement (ablation) | tracking algorithm · PETS, TUD, Town Centre, ETH | MOTA | Improved |
| use DPM detections instead of public detections | full model · MOTChallenge 2D | MOTA: 34.3% -> 49.1% | Improved |
| use DPM detections instead of public detections | full model · PETS2009-S2L2 | MOTA: 51.5% -> 59.7% | Improved |
Closest papers by meaning
- Confidence-Based Data Association and Discriminative Deep Appearance Learning for Robust Online Multi-Object Tracking (TPAMI 2017 · similarity 0.88)
- Probabilistic Tracklet Scoring and Inpainting for Multiple Object Tracking (CVPR 2021 · similarity 0.87)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.87)
- Temporal Coherent Object Flow for Multi-Object Tracking (AAAI 2025 · similarity 0.86)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.85)
- Samba: Synchronized Set-of-Sequences Modeling for Multiple Object Tracking (ICLR 2025 · 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.
