Variational Autoencoder for Nerf dataset

Decoding the latent space into images in the Nerf dataset.

Installation

  pip install -r requirements.txt

Running code

Change directory

  cd /Your path/variational_autoencoder

Create data:

  python dataset/save_vae_data.py

Training model:

  config_path = 'direction path'
  config_name = 'config name'
  HYDRA_FULL_ERROR=1 python run.py --config-path {config_path} --config-name {config_name}

Dataset link

Synthesis Nerf: Link

Objaverse Eval: Link