Enhanced pothole detection system using YOLOX algorithm

Abstract The road is the most commonly used means of transportation and serves as a country’s arteries, so it extremely important to keep roads in good condition. Potholes that happen appear must be repaired Spotting potholes on difficult, especially country like India where stretch millions kilometres across country. Therefore, there need automate identification with high speed real-time precision. YOLOX an object detection algorithm our main goal this article train analyse model for pothole detection. trained dataset results obtained are analysed by calculating accuracy, recall size which then compared other YOLO algorithms. experimental show YOLOX-Nano predicts higher accuracy models while having low computational costs. We were able achieve Average Precision (AP) value 85.6% from training total 7.22 MB. capabilities newly developed have never been tested before paper one first detect using algorithm. research conducted will help reduce costs increase great maintenance.