Prediction of Accidents by Surveillance and Continuous Monitoring of Road Conditions Using Deep Learning Techniques
M.G. Giri, N S Malavika, R. Lakshmi Devi, E. Ranjith Kumar, K. Rohith, Patil Rushikesh
Abstract
India like a developing countries road accident are common and are occurred frequently. Many people lost their life due to accidents, and that are occurred mainly due to driver fault, vehicle faults, and road faults like road damages and government fails to monitor road conditions. In this research we proposed a new technique to detect damages on roads using vehicles of Arial type (UAV) to capture road pictures frequently and to predict incidents we are using deep learning approach. In India maintenance of roads with good condition is a difficult and challenging task. However, collecting road pictures frequently by manual techniques are human interest and also not advisable. In our research we are planned to identify damages on road surfaces, we used versions 4, 5, 7 of YOLO for detecting objects, all these three methods are trained and also tested using dataset, results are evaluated, and our method shows efficient results.