Road Accident Analysis using Structure from Motion and Machine Learning

Frequency of accidents have increased along with an increase in the number of vehicles on the roads. Many people suffer injuries, with many incurring a disability as a result of their injury. This work focuses on analyzing road accidents by using machine learning and videogrammetry techniques with the help of videos and images obtained from unmanned aerial vehicles (UAV). Machine learning model involves classifying accident site images into different causes of accidents. Videogrammetry utilizes structure from motion (SFM) techniques to create a point cloud mesh of the captured video.