Pothole Detection System

Pothole Detection System

Author Name : Prof. Guruprasad Kulkarni, Bibi Fatima

ABSTRACT The aim of this study is to develop a pothole detection system that utilizes a trained model’s knowledge to learn another set of data. Potholes on roads can cause significant damage to vehicles and pose a threat to road safety. Detecting and repairing potholes in a timely manner is crucial for maintaining road infrastructure. This study involves a technique that allows us to use pre-trained deep learning models. As we utilize convolutional network (CNN) pre-trained on images and make it more precise to detect potholes on roads. The study involves several dataset of road images containing both pothole and non-pothole samples collected. In a pre- trained CNN model the next few layers keep the preceding levels frozen and is then trained to perform the specialized task of pothole identification. The project also involves creating an intuitive user interface. that allows users to capture real- time road images and perform pothole detection using the trained model. The interface provides visual feedback on the location and severity of detected potholes, aiding road maintenance crews in identifying areas that require repair.