School Bus Safety Monitoring System Based On Computer Vision

In recent years, the school bus accidents emerge in endlessly in china. The reason behind is the lack of supervision to school bus caused by high monitoring cost. How to improve level of school supervision and reduce the cost of school bus management become the focus of the society.Based on the above background, this paper presents school bus safety monitoring system based on computer vision,which is through the intelligent visual analysis, monitoring the school bus interior safety status, for alerts of abnormal status to ensure school safety driving, while reducing the bus cost of human management.Firstly, this paper uses the face recognition to check the attendance of passengers on the bus.Staff number and status monitoring can be accomplished quickly by this.It also can effectively warn the incoming passengers and staff overload situation to imp-rove school bus safety further.At the same time, this paper uses convolution neural network to extract facial fea-ture points.By extracting facial feature points, with the PERCLOS fatigue detection algorithm, closed time detection and the detection of the movements of the mouth to judge whether the driver is on the abnormal state.This article also presents detection of passengers in the abnormal state, by detect-ing whether passengers move when vehicle driving and warn the movers to ensure safety of passangers.Finally throught comprehensive consideration of the demand of the project and system implementation, this paper decides to use the Jetson TK1 embedded platform which supports CUDA. The algorithm in this paper is implemented in the embedded platform, which has huge value of engineering.