Porting Video-based Crowd Analysis on an Edge Device

Bachelor Thesis, Claudius Kienle, 2021

At large events, video-based analysis tools on crowds are an emerging technology to support local security forces monitor the situation. However, the computations involved are extremely resource-intensive, which hinders on-site use on small terminals without further ado. This work addresses the porting of an existing analytics tool to a terminal device. For this purpose a new design of the existing software with respect to an optimal execution on the end device is developed. The subsequent analysis of the implemented software and a comparison of various implementation options targets the development of a highly efficient, for the end device tailored software.