%0 Conference Proceedings %T Weakly Supervised Multi-Object Tracking and Segmentation %A Idoia Ruiz %A Lorenzo Porzi %A Samuel Rota Bulo %A Peter Kontschieder %A Joan Serrat %B IEEE Winter Conference on Applications of Computer Vision Workshops %D 2021 %F Idoia Ruiz2021 %O ADAS; 600.118; 600.124 %O exported from refbase (http://refbase.cvc.uab.es/show.php?record=3548), last updated on Tue, 23 Nov 2021 12:14:47 +0100 %X We introduce the problem of weakly supervised MultiObject 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 byGrad-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 theobjects. 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. %U http://refbase.cvc.uab.es/files/RPR2021.pdf %P 125-133