Research map / Detection and segmentation / Tracking
Quasi-Dense Similarity Learning for Multiple Object Tracking
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 |
|---|---|---|---|
| Replace sparse ground-truth matching with quasi-dense matching (ablation) | QDTrack · BDD100K tracking val | IDF1: 63.0 to 67.8 | Helped |
| Use multiple positive targets rather than one positive target (ablation) | QDTrack · BDD100K tracking val | IDF1: 66.8% to 67.8% | Helped |
| Replace cosine similarity with bi-softmax (ablation) | QDTrack · BDD100K tracking val | IDF1: 67.8 to 70.0 | Helped |
| Add duplicate removal to matching (ablation) | QDTrack with bi-softmax · BDD100K tracking val | IDF1: 70.0 to 70.1 | Helped |
| Add backdrops to matching candidates (ablation) | QDTrack with bi-softmax and duplicate removal · BDD100K tracking val | IDF1: 70.1 to 71.5 | Helped |
| Add IoU location cues to appearance embeddings (ablation) | QDTrack · BDD100K tracking val | mMOTA: 36.6 to 36.3 | Hurt |
| Add IoU and motion cues to appearance embeddings (ablation) | QDTrack · BDD100K tracking val | mMOTA: 36.6 to 36.4 | Hurt |
| Add IoU and displacement regression to appearance embeddings (ablation) | QDTrack · BDD100K tracking val | mMOTA: 36.6 to 36.4 | Hurt |
| Increase embedding update momentum from 0.6 to 1.0 (ablation) | QDTrack · BDD100K tracking val | IDF1: 71.4 to 70.5 | Hurt |
| Use QDTrack instead of the strongest listed MOT17 private-detector baseline | MOT17 tracking · MOT17 test, private detectors | MOTA: 67.8 to 68.7 | Helped |
| Use QDTrack instead of Yu et al. | BDD100K tracking · BDD100K tracking val | mMOTA: 25.9 to 36.6 | Helped |
| Use QDTrack instead of RetinaTrack | Waymo tracking, Vehicle · Waymo tracking val | MOTA: 44.92 to 55.6 | Helped |
Closest papers by meaning
- SMILEtrack: SiMIlarity LEarning for Occlusion-Aware Multiple Object Tracking (AAAI 2024 · similarity 0.87)
- ODTrack: Online Dense Temporal Token Learning for Visual Tracking (AAAI 2024 · similarity 0.86)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.86)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.85)
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision (NeurIPS 2025 · similarity 0.85)
- SPMTrack: Spatio-Temporal Parameter-Efficient Fine-Tuning with Mixture of Experts for Scalable Visual Tracking (CVPR 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.