SensatUrban

SensatUrban is an extensive, urban-scale 3D point cloud dataset specifically developed for semantic segmentation research. It boasts an impressive nearly three billion richly annotated points, making it a significant resource for the geospatial and machine learning communities. The dataset's sheer volume and detailed annotations are designed to push the boundaries of current point cloud processing techniques.

This dataset is five times larger than existing comparable datasets, covering substantial areas in two major United Kingdom cities. It serves as a foundational resource for researchers, offering a robust platform for developing and evaluating algorithms for semantic segmentation and classification of urban environments. Associated resources include a research paper, blog, video, project page, and a dedicated download portal.

Hosted on GitHub, SensatUrban facilitates access to this valuable labeled-dataset and encourages participation in related research initiatives. The project also hosts the Urban3D Challenge, fostering competition and collaboration in the field of 3D point cloud analysis for urban applications, particularly in areas like infrastructure, buildings, and roads.