Pothole Detection using YOLO v8 Algorithm

• Published in: 2023 9th International Conference on Smart Computing and Communications (ICSCC) pp. 1 - 5
• Main Authors: Bala, Rounak Kumar, Bhattacharyya, Ankit, Roy, Ankan, Roychowdhury, Adrija, Chatterjee, Debraj
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17.08.2023
• Subjects: Accidents; Confusion Matrix; Image recognition; Pothole detection; Precision curve; Predictive models; Public security; Roads; Threshold Value; Training; Urban areas; YOLO
• DOI: 10.1109/ICSCC59169.2023.10335077

With advancing urbanization, roads serve as a most important way of communication. But the increasing potholes on the roads pose a bigger danger towards public safety. In this paper we tend to build a system so as to identify and detect potholes. Here we aim to identify the potholes present on the roads so as to beware of the accidents. In recent years vehicles do come with technologies to detect potholes, but we tend to build an interface to automatically identify and detect the same. As far as we know, most of the fatal accidents happen on roads in India, and majority of them occur due to uneven road terrain caused due to potholes. The most important goal that we hold is to maintain public safety. Conversely, we also wanted to check the performance of the YOLO v8 algorithm, on the pothole dataset, to see how efficiently the algorithm works. Now for the development of the system, we have built a model and used a dataset, containing pothole images and corresponding labels, so as to train the model. We have used the YOLO v8 algorithm to train the model. As a result, our proposed model can successfully identify and recognize potholes after the successful training of the system.