/STTR

[ACCV'22] Fine-Grained Image Style Transfer with Visual Transformers

Primary LanguagePythonMIT LicenseMIT

Install

conda install -c pytorch pytorch torchvision
conda install cython scipy

Test

Download pre-trained model here

https://drive.google.com/file/d/1OsRljnuWBb9BQzFJnidkhwuPKOt6YYx9/view?usp=sharing

Put it under ./checkpoint_model

Please change the images you want to use in the following folders:

inputs/content: a folder to store all content images
inputs/style: a folder to store all style images

Then run

python demo_sttr_image.py

You could see results in outputs/test_outputs/0005