FRACTAL: An Ultra-Large-Scale Aerial Lidar Dataset for 3D Semantic Segmentation of Diverse Landscapes
This dataset addresses a data gap for mapping agencies and researchers needing diverse Lidar benchmarks for 3D semantic segmentation in land monitoring, though it is incremental as it builds on existing data collection methods.
The authors tackled the lack of large annotated aerial Lidar datasets for diverse landscapes by introducing FRACTAL, an ultra-large-scale dataset with 100,000 point clouds covering 250 km² and 7 semantic classes, and provided baseline results using a state-of-the-art 3D model.
Mapping agencies are increasingly adopting Aerial Lidar Scanning (ALS) as a new tool to map buildings and other above-ground structures. Processing ALS data at scale requires efficient point classification methods that perform well over highly diverse territories. Large annotated Lidar datasets are needed to evaluate these classification methods, however, current Lidar benchmarks have restricted scope and often cover a single urban area. To bridge this data gap, we introduce the FRench ALS Clouds from TArgeted Landscapes (FRACTAL) dataset: an ultra-large-scale aerial Lidar dataset made of 100,000 dense point clouds with high quality labels for 7 semantic classes and spanning 250 km$^2$. FRACTAL achieves high spatial and semantic diversity by explicitly sampling rare classes and challenging landscapes from five different regions of France. We describe the data collection, annotation, and curation process of the dataset. We provide baseline semantic segmentation results using a state of the art 3D point cloud classification model. FRACTAL aims to support the development of 3D deep learning approaches for large-scale land monitoring.