Research map / Detection and segmentation / Tracking
MOTR: End-to-End Multiple-Object Tracking with TRansformer
Fangao Zeng, Bin Dong, Yuang Zhang, Tiancai Wang, Xiangyu Zhang, Yichen Wei
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 |
|---|---|---|---|
| add track queries instead of object queries only (ablation) | MOTR baseline · 2DMOT15 train (ablation) | IDF1: 1.2 -> 49.8 | Improved |
| add temporal aggregation network (TAN) to track query baseline (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 37.1 -> 44.9 | Improved |
| add collective average loss (CAL) on top of TAN (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 44.9 -> 53.2 | Improved |
| add TAN on top of CAL (ablation) | MOTR · 2DMOT15 train (ablation) | IDF1: 56.1 -> 70.5 | Improved |
| increase CAL video clip length from 2 to 5 (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 44.9 -> 53.2 | Improved |
| use clip length 4 instead of 3 in CAL (ablation) | MOTR · 2DMOT15 train (ablation) | IDF1: 59.4 -> 64.0 | Mixed |
| raise track query erasing probability from 0.1 to 0.5 (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 53.2 -> 48.5 | Hurt |
| lower track query erasing probability from 0.1 to 5e-2 (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 53.2 -> 49.0 | Hurt |
| raise false positive insertion probability from 0.3 to 0.7 (ablation) | MOTR · 2DMOT15 train (ablation) | IDF1: 70.5 -> 57.7 | Hurt |
| vary entrance threshold from 0.7 to 0.9 (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 52.7 -> 53.1 | No clear effect |
| lower exit threshold from 0.6 to 0.5 (ablation) | MOTR · 2DMOT15 train (ablation) | MOTA: 53.2 -> 53.5 | No clear effect |
| increase exit threshold from 0.6 to 0.7 (ablation) | MOTR · 2DMOT15 train (ablation) | IDF1: 70.5 -> 68.3 | Hurt |
Later papers that compared with this method
13 later accepted papers report MOTRv3 as a baseline in their results tables; 11 of them report a better number on most of the results they share with it.
- SelfMOTR: Revisiting MOTR with Self-Generating Detection Priors (ECCV 2026 · better on 4 of 4 shared results)
- OmniPT: Unleashing the Potential of Large Vision Language Models for Pedestrian Tracking and Understanding (AAAI 2026 · better on 1 of 1 shared result)
- FusionTrack: Collaborative Multi-Object Tracking with Arbitrary Multi-UAVs (ECCV 2026 · better on 1 of 1 shared result)
- More Than Meets the Eye: Enhancing Multi-Object Tracking Even with Prolonged Occlusions (ICML 2025 · better on 2 of 2 shared results)
- LA-MOTR: End-to-End Multi-Object Tracking by Learnable Association (ICCV 2025 · better on 1 of 2 shared results)
- MMOT: The First Challenging Benchmark for Drone-based Multispectral Multi-Object Tracking (NeurIPS 2025 · better on 1 of 1 shared result)
- Beyond MOT: Semantic Multi-Object Tracking (ECCV 2024 · better on 1 of 1 shared result)
- MeMOTR: Long-Term Memory-Augmented Transformer for Multi-Object Tracking (ICCV 2023 · better on 4 of 4 shared results)
5 more not listed.
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- ByteTrack — ByteTrack: Multi-Object Tracking by Associating Every Detection Box (ECCV 2022) · better on 3 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MOTRv2: Bootstrapping End-to-End Multi-Object Tracking by Pretrained Object Detectors (CVPR 2023 · similarity 0.92)
- Looking Beyond Two Frames: End-to-End Multi-Object Tracking Using Spatial and Temporal Transformers (TPAMI 2022 · similarity 0.91)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.89)
- More Than Meets the Eye: Enhancing Multi-Object Tracking Even with Prolonged Occlusions (ICML 2025 · similarity 0.88)
- OVTR: End-to-End Open-Vocabulary Multiple Object Tracking with Transformer (ICLR 2025 · similarity 0.88)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.88)
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.
