Stereo vision and statistical based behaviour prediction of driver

Abstract

The goal of this project is to develop a webeam-based system for monitoring the activities of automobile drivers. As in any system deployed for monitoring driver activities, the primary goal is to distinguish between safe and unsafe driving actions. There is no fixed list of actions that qualifies the unsafe driving behaviors. In general, an activity or an action that reduces a driver's alertness of their surroundings should be classified as unsafe driving behavior. Some examples of unsafe driving behavior include fatigue, talking on a cellular telephone, eating, and adjusting the controls of the dashboard stereo while driving. In this study, we also investigated the relationship between 2D and 3D face and pose recognition.