Weakly Supervised Multi-Object Tracking and Segmentation
This work addresses the challenge of reducing annotation burden for multi-object tracking and segmentation, which is beneficial for researchers and practitioners in computer vision.
This paper tackles the problem of weakly supervised Multi-Object Tracking and Segmentation, which involves joint weakly supervised instance segmentation and multi-object tracking without mask annotations. The proposed method reduces the performance gap on the MOTSP metric between fully supervised and weakly supervised approaches to 12% for cars and 12.7% for pedestrians on the KITTI MOTS benchmark.
We introduce the problem of weakly supervised Multi-Object Tracking and Segmentation, i.e. joint weakly supervised instance segmentation and multi-object tracking, in which we do not provide any kind of mask annotation. To address it, we design a novel synergistic training strategy by taking advantage of multi-task learning, i.e. classification and tracking tasks guide the training of the unsupervised instance segmentation. For that purpose, we extract weak foreground localization information, provided by Grad-CAM heatmaps, to generate a partial ground truth to learn from. Additionally, RGB image level information is employed to refine the mask prediction at the edges of the objects. We evaluate our method on KITTI MOTS, the most representative benchmark for this task, reducing the performance gap on the MOTSP metric between the fully supervised and weakly supervised approach to just 12% and 12.7% for cars and pedestrians, respectively.