Research map / Detection and segmentation / Tracking
Global Tracking Transformers
Xingyi Zhou, Tianwei Yin, Vladlen Koltun, Philipp Krähenbühl
Check 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 pairwise association with global trajectory association | GTR versus QDTrack with the same detector · TAO validation | tracking mAP@0.5: 16.1 -> 20.4 | Improved |
| Use a longer temporal window (ablation) | GTR · MOT17 validation | AssA: 55.8 -> 66.2 | Improved |
| Use a longer temporal window (ablation) | GTR · TAO validation | tracking mAP@0.5: 13.6 -> 22.5; 22.5 -> 22.1 | Mixed |
| Add encoder attention instead of direct dot-product association (ablation) | GTR association head · MOT17 validation | AssA: 63.6 -> 66.2 | Improved |
| Add decoder attention after encoder attention (ablation) | GTR association head · MOT17 validation | AssA: 66.2 -> 64.5 | Hurt |
| Add learned positional embedding (ablation) | GTR · MOT17 validation | AssA: 66.2 -> 65.0 | Hurt |
| Add learned positional and temporal embeddings (ablation) | GTR · MOT17 validation | AssA: 66.2 -> 64.6 | Hurt |
| Increase decoder layers from 1 to 2 (ablation) | GTR · MOT17 validation | AssA: 66.2 -> 65.0 | Hurt |
| Increase encoder layers from 1 to 2 (ablation) | GTR · MOT17 validation | AssA: 66.2 -> 66.0 | No clear effect |
| Add location-based IoU to testing-time association (ablation) | GTR · MOT17 validation | AssA: 63.3 -> 66.2 | Improved |
| Add location-based IoU to testing-time association (ablation) | GTR · TAO validation | tracking mAP@0.5: 22.5 -> 22.5 | No clear effect |
| Train the tracker on the TAO training set (ablation) | GTR · TAO | detection performance | Hurt |
Closest papers by meaning
- TrackFormer: Multi-Object Tracking With Transformers (CVPR 2022 · similarity 0.90)
- TransCenter: Transformers With Dense Representations for Multiple-Object Tracking (TPAMI 2022 · similarity 0.90)
- Multi-Scene Generalized Trajectory Global Graph Solver with Composite Nodes for Multiple Object Tracking (AAAI 2024 · similarity 0.88)
- TGFormer: Transformer with Track Query Group for Multi-Object Tracking (AAAI 2025 · similarity 0.87)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.87)
- OVTR: End-to-End Open-Vocabulary Multiple Object Tracking with Transformer (ICLR 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.
