In this repository, I train a CycleGAN to convert photos to Monet-like paintings and Monet paintings to photos.
To train on custom data, setup the images from two sources in two different directories. Clone this repository. Then run the following command.
python main.py --photos_path one_dir --paintings_path second_dir --run_name give_some_name
The evaluation curves and sample output images will be generated and automatically uploaded in real time to your WandB dashboard. Please create a WandB account to log the results successfully.