Title: Computer Vision and Machine Learning based human behavior analysis in crowded scenes
Promoter: Hichem Sahli

Abel Díaz Berenguer is a Ph.D. candidate at Electronics and Informatics (ETRO) Department under the supervision of Professor Hichem Sahli. He conducts his Ph.D. research on machine learning in the field of computer vision with a focus on developing deep learning algorithms for vision-based human behavior analysis and prediction.

Automatically understanding human behavior is a crucial research topic in the machine learning and computer vision community to enhance the "smart" capabilities of vision-based autonomous systems, such as computer-aided video surveillance. There is an increasing interest to address public safety assisted by autonomous video surveillance systems due to computer-aided video surveillance helps to decrease human labor dedicated to monitoring. Hence, automatic human behavior analysis for automatic visual surveillance has a high impact on public security. Furthermore, the main objective of an automated video surveillance system is to observe, analyze, interpret human behavior, and eventually detect on time unusual activities or events, considered as anomalies. Precisely, Abel's Ph.D. research target human behavior analysis in crowd scenes. The research focuses on developing and incorporating computational mechanisms of computer vision and deep learning algorithms to automatically analyze and predict human behavior in crowded video surveillance footage.

Abel's Ph.D. aims at enhancing and developing data-driven algorithms to address computational problems related to automatic human behavior analysis. It covers the proposal of methodologies and suitable techniques to detect pedestrians within the crowded scene, track the detected pedestrians, predict the trajectory they would follow and detect anomalous human behavior. Using as input the video sequence, these tasks are incorporated into a complete processing approach to automatically understand patterns inside the scene and to detect abnormal human behavior.

More information can be found here.