Road accidents are unpredictable events that take lives of millions of people across the globe. Not all accidents are human error, they can be due to defects in the road infrastructures also. In order to analyze these defects and mark accident hotspots requires precise data. This paper proposes a novel system called RADAR (Road Accident Data Annotation and Reporting System) for extracting crucial data from images of accident scenes captured by a mobile camera, Unmanned Aerial Vehicle (UAV) and surveillance cameras to analyze the causes and hotspots of accidents. The system processes the captured image to extract important information such as the type of collision, weather conditions, the exact location of the accident (which is reverse-geocoded), and the classes and number of vehicles involved in the accident. This data can be used to provide accurate and timely information to emergency services and other stakeholders, and to improve road safety by identifying accident hotspots and implementing targeted interventions. The proposed system is based on deep learning techniques and utilizes state-of-the-art object detection algorithms to accurately identify and classify vehicles and other objects in the accident scene. Experimental results demonstrate the effectiveness and efficiency of the proposed system in accurately extracting important information from accident scenes.
Road Accident Data Annotation and Reporting (RADAR) - A Comprehensive Approach to Automate Accident Images Analysis Using Deep Learning Algorithms
2023-09-01
415285 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Accident reporting and road safety in Mamelodi
British Library Conference Proceedings | 1994
|Road Accident Analysis using Machine Learning
IEEE | 2020
|Online Contents | 2000