Improve command line usage about faceswap HOT 10 CLOSED

Merged #13 and #14 which takes care of the conversion script.

from faceswap.

Also if you have ideas for a better folder structure, feel free to suggest!

from faceswap.

I would propose to keep the different "actions" or steps in their current files (train.py, convert.py, extract.py) and then allow for model data directory names to be used as input/output arguments. Example:

train.py --data-dirs="./data/trump;/data/cage" --model-dir="./models/trump-cage/"

from faceswap.

seems fine to me

from faceswap.

extract.py now has basic command line options for specifying paths and verbose output. I'm going to be looking at convert.py next.

from faceswap.

Updated train.py and accidentally pushed results to upstream. I'm very drunk, so if someone could verify if the scripts are still working, please do so.

Now that all these scripts are using the same basic classes, we can probably consolidate them to a single file so they can be called from a single entrypoint. (like how git has several subcommands, faceswap.py could have extract, train and convert commands) That's my next objective, which I will be looking to tackle tomorrow. (after the hangover passes)

from faceswap.

I pushed also accidentally yesterday and I was not even drunk...

No problem, have a look tomorrow. Me I'll be on the way back to my family...

from faceswap.

If we could manage to create a program with different commands it would be way easier to use indeed.

Something like faceswap extract, faceswap train and faceswap convert

from faceswap.

Thank you, @Yutsa for implementing this functionality.

from faceswap.

This issue should probably be closed and split up into more specific issues since we have some basic command line usage now.

from faceswap.