A partial open-source replication of D4RT: Efficiently Reconstructing Dynamic Scenes One D4RT at a Time (Google DeepMind, CVPR 2026), built using the pretrained VGGT-1B encoder and publicly available datasets.
computer-vision deep-learning transformers pytorch nerf video-understanding depth-estimation camera-pose-estimation dynamic-scenes kubric point-tracking 4d-reconstruction cross-attention vggt d4rt tapvid
-
Updated
Aug 5, 2026 - Python