POTHOLE DETECTION USING DEEP LEARNING

Abstract

Industrialization of transportation system has derived serious accidents that resulted in thousands of deaths. To solve the problem, vision-based pothole detection for advanced driver assistance system has been researched. In this study, we provide experimentations of pothole detection and localization in on-road environment using deep learning. In India as poor quality of construction materials get used in road drainage system construction. Due to the above problems, roads get damaged early and potholes appear on the roads which cause accidents. According to a report submitted by the Ministry of Road Transport and Highways transport research wing New Delhi in 2022, approximately 1,55,622 accident deaths happen in India. This project proposed a deep learning-based model that can detect potholes early using images and videos which can reduce the chances of an accidents. Detailed real-time performance comparison of state-of-the-art deep learning models and object detection frameworks (YOLOv1, YOLOv2, YOLOv3, YOLOv4, Tiny-YOLOv4, YOLOv5, and SSD-mobilenetv2) for pothole detection is presented. The experimentation is performed on an image dataset with pothole in diverse road conditions and illumination variations as well as on real-time video captured through a moving vehicle. The Tiny-YOLOv4, YOLOv4, and YOLOv5 evince the highest mean average precision of 80.04%, 85.48%, and 95%, respectively, on the image set, thus proving the strength of the proposed approach for pothole detection and deployed on OAK-D for real-time detection. The study corroborated Tiny-YOLOv4 as the befitted model for real-time pothole detection with 90% detection accuracy and 31.76 FPS. For developing this project, we have used YoloV4-Tiny and OpenCV.