Research And Implementation Of Vehicle Accident Detection System Based On Deep Learning

Nowadays,with the rapid development of information technology,people pay more and more attention to the research and application of artificial intelligence,especially in image processing,common traffic accident detection system is a large number of the use of image processing technology.For the existing traffic accident detection systems,most of them do not take into account the impact of low light environment on the accuracy of vehicle accident recognition,this makes it difficult for the existing traffic accident detection system to accurately detect the vehicle accident in low light conditions,which has a great impact on urban traffic conditions.At present,there are many processing methods for low-light image and video,but most of the methods based on neural network are too complex,the network model is too complex,and the training process is complicated,this limits the availability and Operability of the entire system in a real world scenario.Firstly,the two existing neural network models Glad Net and Dense Net are improved to realize vehicle accident detection in low light condition,in order to improve the detection system in extreme conditions of the effective detection rate of vehicle accidents.The vehicle accident detection system studied in this subject is mainly based on deep learning theory and technology,which can improve the accuracy of vehicle accident detection in low-light environments.In the vehicle accident detection system,it mainly includes: the user starts the Raspberry Party to collect video streams at the scene of the traffic accident,and enhances the brightness of the video;and detects the vehicle accident on the video with the enhanced brightness.In order to improve the accuracy of the detection system for vehicle accident detection,this paper introduces mechanisms and methods such as noise reduction and detail recovery in the preprocessing stage of traffic accident detection.After that,in the process of vehicle accident detection,attention in computer vision is introduced.and GRU mechanism.Firstly,the improved Glad Net neural network is used to enhance the brightness of traffic accident video data in low light environment,and then the improved Dense Net network is used to detect vehicle accidents,and finally statistical results such as accident rate are obtained.The system adopts the C/S architecture,and the user can conveniently realize the detection of vehicle accidents through the system interface.Through theoretical analysis and a large number of experiments,it can be seen that the vehicle accident detection system in low light environment based on deep learning has good performance in noise reduction,detail recovery,and vehicle accident detection.