Research map / Detection and segmentation / Tracking
Track To Detect and Segment: An Online Multi-Object Tracker
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 |
|---|---|---|---|
| Add CVA to the baseline tracker (ablation) | Baseline tracker · MOT17 validation | IDF1: 59.5 -> 66.7 | Helped |
| Add MFW using feature difference as motion cue (ablation) | Baseline tracker · MOT17 validation | FN: 31.0% -> 29.5% | Helped |
| Add CVA and MFW, using CVA offsets to guide feature propagation (ablation) | TraDeS · MOT17 validation | MOTA: 64.8 -> 68.2 | Helped |
| Replace CVA with CE embedding (ablation) | Baseline with DA-Round (ii) · MOT17 validation | MOTA: 66.5 -> 64.5 | Hurt |
| Use CVA instead of CE embedding (ablation) | Baseline with DA-Round (ii) · MOT17 validation | IDF1: 64.3 -> 66.7 | Helped |
| Add DA-Round (ii) long-term association (ablation) | Baseline+CVA · MOT17 validation | MOTA: 65.5 -> 66.5 | Helped |
| Use feature difference and CVA tracking offset instead of feature difference alone (ablation) | TraDeS MFW · MOT17 validation | MOTA: 67.1 -> 68.2 | Helped |
| Increase aggregated history from one to two previous features (ablation) | TraDeS MFW · MOT17 validation | MOTA: 67.8 -> 68.2 | Helped |
| Increase aggregated history from two to three previous features (ablation) | TraDeS MFW · MOT17 validation | MOTA: 68.2 -> 67.5 | Hurt |
| Use TraDeS rather than CenterTrack | 2D multi-object tracking · MOT17 test | MOTA: 67.3 -> 69.1 | Helped |
| Use TraDeS rather than CenterTrack | 3D multi-object tracking · nuScenes validation, All | AMOTA: 6.8 -> 11.8 | Helped |
| Use TraDeS rather than TrackR-CNN | Instance segmentation tracking · MOTS test | sMOTSA: 40.6 -> 50.8 | Helped |
Closest papers by meaning
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.88)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.87)
- ADA-Track: End-to-End Multi-Camera 3D Multi-Object Tracking with Alternating Detection and Association (CVPR 2024 · similarity 0.86)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.86)
- S2-Track: A Simple yet Strong Approach for End-to-End 3D Multi-Object Tracking (ICML 2025 · similarity 0.85)
- Hybrid-SORT: Weak Cues Matter for Online Multi-Object Tracking (AAAI 2024 · 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.