2014 - 2016
Due to the existence of a large volume of data obtained from surveillance cameras, the understanding and automatic interpretation of activities performed by humans in videos is of great interest in order to assist the task of security agents. Automatic monitoring of monitored environments will enable the development of new technologies, such as accident prevention systems in busy environments and systems capable of recognizing suspicious activities with the aim of preventing crimes. Among other problems to be solved, pedestrian detection is essential so that the enormous amount of visual data captured from surveillance cameras is reduced to a volume that can be managed by current computing systems so that the activities being performed by agents present at the scene can be analyzed. In this way, this project proposes the development of methods for pedestrian detection with the aim of reducing the computational cost and maintaining the accuracy obtained by detectors that obtain good results but have a high computational cost.