I am reimplementing AdaIn for the OAK-1. You can see progress at my blog. TLDR, I have gotten the model training, and the images look decent for now. Next I will shrink the model for the OAK-1.
You can run the code like so:
python3 main.py --crop "run description 1 2 3"
Here are some results (style, content, stylized):
