Extract with dlib-cnn doesn't work on macOS

[helloall1900]

Note: Please only report bugs in this repository. Just because you are getting an error message does not automatically mean you have discovered a bug. If you don't have a lot of experience with this type of project, or if you need for setup help and other issues in using the faceswap tool, please refer to the faceswap-playground instead. The faceswap-playground is also an excellent place to ask questions and submit feedback.

Expected behavior

Describe, in some detail, what you are trying to do and what the output is that you expect from the program.

Actual behavior

Describe, in some detail, what the program does instead. Be sure to include any error message or screenshots.

Steps to reproduce

Describe, in some detail, the steps you tried that resulted in the behavior described above.

Other relevant information

• Command lined used (if not specified in steps to reproduce): faceswap.py ...
• Operating system and version: Windows, macOS, Linux
• Python version: 2.7, 3.5, 3.6.4, ...
• Faceswap version: commit hash or version number
• Faceswap method: CPU/GPU
• Other related issues: Face state detection & categorization? #123, Small error in lib/cli.py get_faces(self, image)  #124...
• ... (for example, installed packages that you can see with pip freeze)

Extract with dnn-lib on mac OS X 10.13.5 failed:

Failed to extract from image: NoneType is not iterable

GPU: NVIDIA GTX 1080, 8GB
CUDA: 9.1
Python: 3.6.5

Reason is NVML library is not supported on macOS, so GPUStats.initialize() in gpu_stats.py fail with pynvml.NVMLError_LibraryNotFound, self.handles = None, get_free() gets Exception: "NoneType is not iterable" when extracting with dlib-cnn, no face will be dectected.
Can we work around it?

[torzdf]

NVML is kind of integral to extract. It may be that we may have to add some kind of manual config for macOS, but unfortunately it's not a priority right now. I will look into it when I have a chance (or someone else can raise a PR).

In the meantime you can rollback to the last pre-nvml build (https://github.com/deepfakes/faceswap/tree/f90cd92ec36a0860c58a75631f762069419149de)

[torzdf]

implemented in #510