HODome (NeuralDome)

HODome Dataset (NeuralDome, CVPR 2023) HODome is a dense 76-view dome capture of human–object interaction: 10 subjects interacting with 23 objects. This repository hosts the annotations (body-model reconstructions, object 6-DoF, calibration, scanned object templates). The 76-view 4K videos are large and distributed separately (see Videos below). Paper: NeuralDome (CVPR 2023) Project page: Toolbox (download / process / visualize):… See the full description on the dataset page:

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