Classified TLS and SLAM Point Clouds for Semantic Analysis in Built Environments – Uniud Rizzi Campus (Courtyard)

This dataset, entitled “Uniud Rizzi Campus (Courtyard)”, contains semantically classified three-dimensional point clouds acquired in an outdoor built environment and processed using different mapping technologies. It was created to support comparative analyses of semantic classification performance on point clouds generated by Terrestrial Laser Scanning (TLS) and by multiple Simultaneous Localization and Mapping (SLAM) algorithms. The dataset includes one TLS-derived reference point cloud and four SLAM-derived point clouds reconstructed from the same raw mobile robotics mapping data using different SLAM approaches (RTAB-Map, LIO-SAM, DLO, and LeGO-LOAM). All point clouds are provided with manual semantic annotations and predefined training and test splits corresponding to those adopted in the associated experimental study. Detailed information on dataset structure and usage is provided in the accompanying README.md file. When using this dataset, please cite the associated publication: Matellon, A., Maset, E., Beinat, A., & Visintini, D. (2026). How the Choice of SLAM Algorithm Impacts Point Cloud Classification: An Experimental Evaluation. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLVIII-2/W12-2026, 295-302, https://doi.org/10.5194/isprs-archives-XLVIII-2-W12-2026-295-2026 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - The point clouds included in this dataset are derived from the data acquisition campaign originally presented in: Tiozzo Fasiolo, D., Scalera, L., Maset, E., 2023. Comparing LiDAR and IMU-based SLAM approaches for 3D robotic mapping. Robotica, 41(9), 2588–2604, https://doi.org/10.1017/S026357472300053X While the original publication focused on the comparison of SLAM approaches for 3D mapping, the present dataset provides manually classified point clouds and predefined training and test splits, enabling further research on semantic classification and downstream analysis. Users interested in the raw sensor data are referred to the original publication and to the dataset available at: https://github.com/diegotiozzo21/mobile-robotics-uniud-datasets