Deep Learning-based Real-time Traffic Accident Type and Fault Information Provision Service

Determining the percentage of negligence between the parties in the event of road traffic accidents is a significant problem. In order to provide users with more accurate criteria for determining the percentage of negligence, several companies are providing services. However, services currently available are limited to immediate use at the scene of an accident. Generally, the service that determines the percentage of negligence can be used after all accident handling procedures have been completed. This paper provides a real-time traffic accident type and fault rate information provision service utilizing a deep learning-based predictive model to overcome these limitations. Users can immediately identify accident types and fault information by taking pictures at the accident site and check actual precedents of the same accident type. Users will be able to use the service to more accurately and reliably determine the percentage of negligence and handle incidents.