Road Accidents Severity Prediction and Brake Lights Detection Using Decision Level Fusion of Ann And Yolov8

Abstract

Road accidents continue to be a global concern, causing immense loss of life and property. Traditional approaches to accident prevention have limitations in accurately detecting and predicting accidents in real-time. The proposed integrated Machine Learning and Deep Learning model for the avoidance of road accidents and to achieve accurate accident detection and proactive prevention. The Combined Road Accident Dataset (CRAD) is used for important feature extraction and severity prediction. The Rear End Collision Dataset (RECD) is used for brake light detection. The Machine Learning model is trained using Artificial Neural Network (ANN) and Deep Learning model is trained by novel You Look Only Once Version 8 (YOLOv8) algorithm. An ensemble of ANN and YOLOv8 algorithms is evolved into a Decision Level Fusion model that assists the driver. newlineThe proposed model, designed to develop an Ensemble Model that assists drivers in avoiding road accidents, comprises three models. The Machine Learning based model predicts the risk level of road accidents. To prevent rear-end collisions, the Deep Learning based model identifies the brake lights of front vehicles. The Ensemble Model is provided to assist the driver and help prevent road accidents. An Ensemble model of a road accident avoidance system integrates Machine Learning and Deep Learning based models newline

Description

Keywords

Citation

item.page.endorsement

item.page.review

item.page.supplemented

item.page.referenced