Downloading the face detection CNN. Please wait......

[jackyxie511]

Describe the bug when I use "bash run.sh",it has always been the following situation. How to do it？

bash run.sh Creating virtual camera /dev/video9 (sudo privelege required) [1608780572.440412] Loading Predictor /home/lab134/xq/xq0914-avatarify-master/avatarify/afy/predictor_local.py:34: YAMLLoadWarning: calling yaml.load() without Loader=... is deprecated, as the default Loader is unsafe. Please read https://msg.pyyaml.org/load for full details. config = yaml.load(f) Downloading the face detection CNN. Please wait...

[ochen1]

Please wait...

I think waiting is what you should do. Does the program not download? Is there a problem where the program exits?

[JohanAR]

Did it finally download the CNN?

[benoit-cty]

Hello, Downloading is very long and, at least in my case, it re-download it everytime I run avatarify. How to make it download only the first time ? I use "bash run.sh --docker" to run it.

[JohanAR]

Is the current directory writable? I don't remember exactly how it's supposed to work, and I can't test right now because it seems like I broke some dependencies when upgrading my OS

[benoit-cty]

Thanks for your answer ! I works for me but I think it's a problem with the Docker implementation : the container may be recreated at each run. Do you know the folder where the file is downloaded ? So we could use a docker volume or a mount folder to store them.

[JohanAR]

The container is definitely recreated each run, this is by design (not by me though). I think the following mapping is supposed to put the model in the current directory: $PWD:/root/.torch/models .But perhaps the PyTorch data location has been changed in later versions, so that it saves it to a non-mapped folder?