IP2RG-PAVEMENT-TYPE-DATASET

The pavement image dataset used in this study was collected using the LUKUN3D intelligent vehicle, a 3D triangulation system developed by the IP2RG research group. The system captures both 2D intensity images and elevation data with a resolution of 1600 × 1000 pixels. Each image represents a pavement section approximately 2.98 meters wide and 3.3 meters long. The dataset includes samples from asphalt and concrete pavements across multiple provinces, covering diverse environmental and traffic conditions. Example images highlight clear differences in surface texture between asphalt and concrete.
The dataset is currently used for pavement type classification, as detailed in the paper Automated Image-level Pavement Type Recognition on Cross-regional Data Using a Multi-feature Fusion Network. The IP2RG team plans to release updated versions of the dataset, with a particular focus on reducing noise in asphalt pavement samples.