MOTR: End-to-End Multiple-Object Tracking with Transformer

7 May 2021  ·  Fangao Zeng, Bin Dong, Yuang Zhang, Tiancai Wang, Xiangyu Zhang, Yichen Wei ·

Temporal modeling of objects is a key challenge in multiple-object tracking (MOT). Existing methods track by associating detections through motion-based and appearance-based similarity heuristics. The post-processing nature of association prevents end-to-end exploitation of temporal variations in video sequence. In this paper, we propose MOTR, which extends DETR and introduces ``track query'' to model the tracked instances in the entire video. Track query is transferred and updated frame-by-frame to perform iterative prediction over time. We propose tracklet-aware label assignment to train track queries and newborn object queries. We further propose temporal aggregation network and collective average loss to enhance temporal relation modeling. Experimental results on DanceTrack show that MOTR significantly outperforms state-of-the-art method, ByteTrack by 6.5% on HOTA metric. On MOT17, MOTR outperforms our concurrent works, TrackFormer and TransTrack, on association performance. MOTR can serve as a stronger baseline for future research on temporal modeling and Transformer-based trackers.Code is available at https://github.com/megvii-model/MOTR.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Multi-Object Tracking DanceTrack MOTR HOTA 48.4 # 2
DetA 71.8 # 5
AssA 32.7 # 2
MOTA 79.2 # 8
IDF1 46.1 # 3
Multi-Object Tracking MOT16 MOTR MOTA 66.8 # 10
IDF1 67.0 # 6
Multi-Object Tracking MOT17 MOTR MOTA 67.4 # 13
IDF1 67.0 # 10

Methods