SpatialLM-Llama-1B

SpatialLM-Llama-1B is a 1-billion-parameter language model fine-tuned for 3D spatial scene understanding, converting point-cloud inputs from monocular video, RGB-D, or LiDAR into structured outputs including walls, doors, windows, furniture labels, and oriented 3D bounding boxes. It is designed to support robotics perception, embodied AI navigation, and spatial reasoning tasks. Source: supabase.