Recently, some weakly supervised multi-object tracking (MOT) methods learn identity embedding features with pseudo identity labels rather than the high-cost manual ones. However, these pseudo identity labels may contain many false or missing identities, which adversely affect the optimization of tracking networks, resulting in interrupted trajectories of occluded targets. To effectively reconnect the interrupted trajectories caused by noisy pseudo labels, we propose a novel weakly supervised MOT method based on a Trajectory-Reconnecting Transformer (TRTMOT). TRT-MOT performs feature decoupling to extract discriminative embedding features for reconnecting trajectories of occluded targets. Experimental results show that TRTMOT outperforms previous weakly supervised MOT methods by at least +3.6 and +5.6 on MOTA for the MOT17 and MOT20 datasets, respectively.

How to generate robust trajectories of multiple objects without using any manual identity annotation? Recently, identity embedding features from Re-ID models are adopted to associate targets into trajectories. However, most previous methods equipped with embedding features heavily rely on manual identity annotations, which bring a high cost for the multi-object tracking (MOT) task. To address the above problem, we present an unsupervised embedding and association network (UEANet) for learning discriminative embedding features with pseudo identity labels. Specifically, we firstly generate the pseudo identity labels by adopting a Kalman filter tracker to associate multiple targets into trajectories and assign a unique identity label to each trajectory. Secondly, we train the transformer-based identity embedding branch and MLP-based data association branch of UEANet with these pseudo labels, and UEANet extracts branch-dependent features for the unsupervised MOT task. Experimental results show that UEANet confirms the outstanding ability to suppress IDS and achieves comparable performance compared with state-of-the-art methods on three MOT datasets.