This dataset is proposed by FreeGave.
The structure of the dataset is as:
DynObjects
| - data
| | - pen1:
| | | - train: serves as training data
| | | - val: used for evaluating novel view interpolation
| | | - test: used for evaluating future extrapolation
| | | - transforms_train.json: camera poses and other meta informations for training set
| | | - transforms_val.json: camera poses and other meta informations for novel view… See the full description on the dataset page:
https://huggingface.co/datasets/scintigimcki/FreeGave-GoPro.
