Fragmented Boulder

Research on fragmented boulders focuses on automating their identification, characterization, and analysis across diverse contexts, from planetary surfaces to terrestrial quarries. Current efforts leverage deep learning, particularly convolutional neural networks (CNNs) and graph neural networks (GNNs), along with instance segmentation models like Mask R-CNN, to process high-resolution images and 3D scans, extracting information on size, shape, and distribution. These advancements improve efficiency and accuracy in applications ranging from resource management in mining to hazard detection for spacecraft landings and furthering our understanding of geological processes. The resulting datasets and improved analytical tools are significantly advancing our ability to study fragmented rock formations across various scales and environments.