HIGHWAY ACCIDENT DETECTION AND NOTIFICATION USING MACHINE LEARNING

Study shows that accidents in India are one of the major causes of deaths and accounts for about 64.4% of the total deaths in India, according to road transport and highways ministry report on 'Road Accidents in India,2018' and this number is ever increasing. One of the main reasons for this number is no timely availability of help. There are many proposed systems to detect these accidents and notify hospitals. Most of them include sensing objects which might cause accident or uses G-force to detect them. This project is about a system that can be used to detect accidents using Image Processing. The system will collect necessary information from CCTV cameras and process that information using Machine learning tools to detect possible accidents. Machine learning algorithms have shown success on distinguishing abnormal behaviors from normal ones. This project aims to analyze vehicles and their state to conclude a scenario as a potential accident. The dataset consists of 800 images, in that 400 for mild crash and 400 for severe crash out of which 200 images for training, and 200 for testing and validating purpose. .The algorithm showed with the performance of 93% accuracy. Results show that clustering algorithms can successfully detect accidents.