SAFE DRIVE: ADVANCED POTHOLE DETECTION SYSTEM FOR ENHANCED ROAD SAFETY IN RAINY CONDITIONS

Computer Engineering

In rainy weather, the presence of potholes on road surfaces constitutes a formidable hazard to vehicular safety. These potholes, formed through the synergistic effects of wear and tear alongside rain-induced erosion, precipitate accidents, inflict vehicle damage, and cause significant traffic disruptions. This project seeks to harness the power of deep learning techniques to develop an advanced object detection system with the specific aim of identifying potholes obscured beneath accumulated rainwater. The inherent challenge in this endeavor is the paucity of image data depicting potholes under turbid rainy water conditions. Current imaging technologies fall short in acquiring clear images in such turbid environments, necessitating an innovative approach. To overcome this limitation, we propose the generation of synthetic images based on sophisticated physical models of underwater scenes. These generated images will serve as the training data for a YOLOv8 model, enabling real-time detection of submerged potholes. The endeavor will not only focus on creating a robust training dataset but will also involve a meticulous comparison of the detection capabilities of the YOLO model when trained with conventional images versus those generated for underwater conditions. This project, therefore, represents a confluence of advanced machine learning, image generation techniques, and practical application in road safety. By enhancing the real-time detection capabilities of potholes under rainy conditions, we aim to mitigate the risks posed by these hazards, thereby contributing to safer driving experiences and reducing the incidence of traffic accidents and vehicle damage. The ultimate goal is to create a reliable and efficient system that can be deployed in real-world scenarios, ensuring enhanced road safety even in adverse weather conditions.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.