Traffic Accident Analysis and Prediction Model Based on Highway Pedestrian Prediction Using Deep Learning Model

Abstract

Due to the high number of deaths, injuries, and fatalities as well as the enormous financial losses they cause every year, traffic accidents rank among the world's most serious worries. Road accidents can be caused by a variety of circumstances. It might be able to take action to lessen the severity and extent of the effects if these elements are better understood and forecast. In order to analyse the data, uncover hidden patterns, forecast the accident severity, and compile the information in an understandable manner, machine learning techniques are employed. In this research the novel deep learning model has been proposed for highway traffic accident analysis based on pedestrian detection in image analysis. here the highway traffic images has been collected and analysed for pedestrian detection using histogram residual Hopfield convolutional neural networks (HRHCNN) and feature selected using markov belief gradient discriminant analysis (MBGDA). the segmented selected features shows the detected pedestrian in highway traffic accident. In simulation results the various pedestrian dataset has been analysed in terms of  AUC, F1 score, MCC, ATA, recall. The proposed technique achieved Average F-1 score was 82%, recall was 90%, AUC was 85%, ATA was 98%, and MCC was 96%.