Pothole Detection and Filling System using Image processing and Machine Learning

Potholes pose a significant risk to vehicles and pedestrians, necessitating prompt action to be addressed. In this project, a strategy is incorporated for pothole detection, utilizing the YOLOv8 object detection algorithm and trained on datasets collected from metropolitan roads exhibiting potholes. The transfer-learning is used to train the respective model, achieving an average precision of 0.82 and a recall of 0.79 upon testing with a dedicated dataset. Furthermore, the model is integrated with hardware components such as Raspberry Pi and Arduino Mega. The Raspberry Pi, equipped with a PiCam serves as a processing unit for real-time analysis of road conditions, while the Arduino Mega facilitates control over the vehicle via serial communication. This integration not only improves speed and accuracy but also reduces human labor, maintenance costs and enhances road safety.