“Let’s get ready to bundle!”: Crowd-level Human Keypoint Tracking

Published in Proceedings of the 2020 Joint Workshop of Fraunhofer IOSB and Institute for Anthropomatics, Vision and Fusion Laboratory, 2021

This work examines the suitability of a state-of-the-art human pose trackingmethod for application within surveillance scenarios and focuses on publicplaces in urban areas that tend to suffer from crowdedness, such as city centers.Starting with a short introduction to motivate keypoint tracking in surveillanceapplications, this report will present details about the adapted method, whichfollows an LSTM-based approach. Afterwards, different changes that had tobe incorporated in order to successfully apply the given method to our targetsetting will be presented. Finally, various experiments will show how the chosenmethod performs, based on experiments with simulated data.