Intelligent Vehicle Collision Warning System Based on a Deep Learning Approach

Developing vehicle collision warning systems on mobile devices aiming to alert drivers about driving environments, and possible collision with other vehicles has become more and more popular. In these systems, robust and reliable vehicle detection is a critical step. This paper presents a vision-based vehicle detection system using a deep learning approach on mobile platforms. Our focus is on the mobile system with camera which is mounted on the vehicle. Integrating detection with tracking is also discussed to illustrate the benefits of deep learning for vehicle detection. Finally, we present the high efficient experimental results based on mobile device mounted on a car.