SPRDD
Year: 2026
Type: dataset
Abstract: SPRDD (Strip Patch Road Damage Dataset) is a public asphalt pavement image dataset designed for robust strip patch detection. It contains 3,500 top-down road surface images collected during highway maintenance inspections in Yunnan Province, China. The dataset provides 2,777 strip patch annotations and 1,866 annotations of four high-frequency pseudo-distress samples: road markings, water stains, shadows, and cracks. All annotations are released in Pascal VOC, YOLO, and COCO formats. SPRDD is the public dataset to combine strip patch labels with dedicated pseudo-distress annotations, enabling systematic evaluation of false-positive robustness. It supports the development and benchmarking of deep learning models for practical field inspection applications.
Authors: Li, Lecheng, Liu, Guoxi, Huang, Bi, Yang, Jing, Dai, Fei
Language: English
Subfield: Civil and Structural Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204235447