DanceNet3D

DanceNet3D A large-scale synchronized multi-view dance dataset captured with 28-29 calibrated cameras at 800×1280 resolution. The dataset contains 46 dance sequences across 3 recording sessions, totaling 36,319 frames with per-frame camera calibration (COLMAP), foreground masks, optional color correction LUTs, and per-frame sparse point cloud reconstructions. Dataset Summary Session Sequences Cameras Frames Size s4 10 29 7,723 ~35 GB s5 23 29 17,928… See the full description on the dataset page:

Source: Hugging Face Hub (nyuvideolab/danceNet3D). Metadata imported from the dataset’s Hub tags.