Real-Time Pothole Detection: A YOLO-Based Approach for Safer Roads

Computer Science and Engineering

Potholes are a major hazard on highways since they are the primary cause of accidents. Preventing accidents requires early detection and correction. These traffic dangers are a difficulty for drivers, especially during the rainy season. To solve this problem, a variety of methods have been used, including vibration-based sensors and manual examination. Fixing potholes is an essential part of maintaining roads, and regulating organizations always struggle to keep an eye on the state of the pavement. An easier way to detect potholes is required because the current approach requires tedious manual picture processing. But there are disadvantages to the current approaches, namely high prices and dangers associated with detection. A non-invasive method using the YOLO (You Only Look Once) algorithm has been proposed to get around these restrictions. Convolutional neural networks (CNNs) are used by YOLO, a real-time object identification system, to identify and classify potholes in photographs. This deep learning project also tries to address pothole issues encountered by self-driving or autonomous automobiles. Real-time pothole identification is provided by this method using emphasized visual clues. By simultaneously predicting bounding boxes and object classes, the CNN-based system improves accuracy and responsiveness. For training, testing, and validation purposes, a dataset including 720x720 pixel quality photos that capture various pothole scenarios in natural road conditions was employed. Using CNN-based object identification algorithms, this unique approach seeks to enable real-time pothole detection and highlighting. This study offers a thorough assessment of YOLOv8, an object identification model, with a focus on identifying potholes and other road hazards.

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