Deep Learning and Approach for Tracking People's Movements in a Video

Jemai Bornia, Ali Frihida, Olivier Debauche, Sidi Ahmed Mahmoudi, Pierre Manneback

Abstract: Everyday, a large amount of data is produced thanks to technological advances in the field of multimedia, associated with the generalization of their use in many applications. The need to keep control over this content, in terms of data analysis, classification, accurate AI (Artificial Intelligence) algorithms are required to perform this task efficiently and quickly. In this article, we propose an approach using deep learning technologies for the analysis of movement in video sequences. The suggested approach uses images from video splitting to detect objects / entities present and store their descriptions in a standard XML file. As result, we provide a Deep Learning algorithm using TensorFlow for tracking motion and animated entities in video sequences.