The dataset is a new dataset for training and evaluating interpretation methods for large, uncurated aerial LiDAR scans. The research team used data from the French Survey and Mapping Agency associated with the LiDAR-HD project. The dataset contains 7 scenes covering more than 7.7 square kilometers and a total of 98 million 3D points, with different content and complexity, such as densely populated areas, forests, or complex industrial facilities.

This dataset is in the paperLearnable Earth Parser: Discovering 3D Prototypes in Aerial Scans", which has been accepted by CVPR 2024.