DL4Lidar model weights - Mask R-CNN Based Model Weights for Atmospheric Semantic Segmentation Using Lidar Range-Corrected Signal Measurements
Year: 2026
Type: dataset
Abstract: This record contains the trained model weights for the Mask R-CNN architecture optimized for atmospheric semantic segmentation (aerosol and cloud detection) in ground-based lidar range-corrected signal images. These weights correspond to the deep learning models presented, validated, and evaluated in Canella-Ortiz et al. (2026). They were trained using the companion DL4Lidar dataset (available at https://zenodo.org/records/20156234), which contains data collected from the ALHAMBRA and MULHACEN lidar instruments at the UGR station (Granada, Spain). For the complete source code, training workflows, configuration files, and inference scripts associated with these weights, please visit the official GitHub repository at https://github.com/acanort/DL4Lidar. If you use these model weights, the dataset, or the repository code in your research, please cite the corresponding journal article: A. Canella-Ortiz, S. Tabik, S. Fernández-Carvelo, O. Rodríguez-Navarro, L. Alados-Arboledas and A. del Águila, "Generalizable Deep Learning for Aerosol and Cloud Segmentation in Ground-Based and Spaceborne Lidar Observations", IEEE Transactions on Geoscience and Remote Sensing, DOI: 10.1109/TGRS.2026.3707646.
Institution: Universidad de Granada
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7169100718