first-order-model study notes (1): environment configuration

Since the library that the project relies on is relatively old, after downloading the source code, it is best to establish an independent Interpreter. The libraries that need to be installed include face-alignment. The numpy version can use 1.20.1. The package name corresponding to skimage is a bit difficult to find. It is called scikit-image, which can be used 0.18.1. The torch version configured on the notebook can use 1.60+cpu.

Before running demo.py, you need to set the parameters, such as: --config config/vox-256.yaml --driving_video resources/videos/00.mp4 --source_image resources/images/01.jpg --checkpoint checkpoints/vox-cpk .pth.tar --relative --adapt_scale

If you want to use cpu for testing, the parameters are: --config config/vox-256.yaml --driving_video resources/videos/00.mp4 --source_image resources/images/01.jpg --checkpoint checkpoints/vox-cpk.pth .tar --relative --adapt_scale --cpu

If an error is reported:

imageio.core.fetching.NeedDownloadError: Need ffmpeg exe. You can obtain it with either:
  - install using conda: conda install ffmpeg -c conda-forge
  - download using the command: imageio_download_bin ffmpeg
  - download by calling (in Python): imageio.plugins.ffmpeg.download()

try:

>>