A YOLOv11-Based Robust, Scalable, and Efficient Pothole Detection Approach

Road safety and infrastructure upkeep are impacted by pothole detection, which is a crucial problem in intelligent transportation systems. This study uses YOLOv11, a cutting-edge deep learning model, to suggest a reliable and effective pothole detection technique. By utilizing cutting-edge features like C3K2 blocks and Spatial Pyramid Pooling Fast (SPFF), YOLOv11 maintains real-time performance while improving detection accuracy, particularly for tiny potholes. Using a wide range of conditions and a diverse dataset of over 3000 images, the model’s accuracy is (92.3 percent), with high sensitivity (90.5 percent) and specificity (89.1 percent). It shows promise for real-time applications by functioning well in difficult situations like dim lighting and rainy weather. In order to increase robustness and practical applicability, future research will concentrate on dataset expansion and depth estimation for severity assessment.