Pavdistress: A Dataset of Selected Images for the Detection and Recognition of Asphalt Pavement Distress Using Deep Learning

The images in the Pavdistress dataset were captured by a road inspection vehicle in the Eastern region of Morocco. This dataset comprises 8,696 images of pavement deterioration. Created from data collected by inspection vehicles, the Pavdistress dataset serves as a benchmark for detecting road damage. It is particularly useful for local authorities and road managers for cost-effectively monitoring road conditions. Five types of road damage are listed in the dataset: edge cracks, longitudinal cracks, potholes, transverse cracks, and road markings. Researchers can use this dataset as a benchmark to evaluate the performance of different algorithms in solving similar problems, such as image classification and object detection.