Eliminating Exposure Bias and Loss-Evaluation Mismatch in Multiple Object Tracking
Identity Switching remains one of the main difficulties Multiple Object Tracking (MOT) algorithms have to deal with. Many state-of-the-art approaches now use sequence models to solve this problem but their training can be affected by biases that decrease their efficiency.
We introduce a new training procedure that confronts the algorithm to its own mistakes while explicitly attempting to minimize the number of switches, which results in better training. We propose an iterative scheme of building a rich training set and using it to learn a scoring function that is an explicit proxy for the target tracking metric. Whether using only simple geometric features or more sophisticated ones that also take appearance into account, our approach outperforms the state-of-the-art on several MOT benchmarks.
Results
First video presents examples of the obtained tracking results on DukeMTMC dataset.
Embed of video is only possible from Mediaspace, Vimeo or Youtube
Second video presents examples of the obtained tracking results on MOT17 dataset.
Embed of video is only possible from Mediaspace, Vimeo or Youtube
Third video presents tracking results obtained by using only geometric features.
Embed of video is only possible from Mediaspace, Vimeo or Youtube
Please note that the publication lists from Infoscience integrated into the EPFL website, lab or people pages are frozen following the launch of the new version of platform. The owners of these pages are invited to recreate their publication list from Infoscience. For any assistance, please consult the Infoscience help or contact support.