search

winfried-ripken / wise

Code for paper "WISE: Whitebox Image Stylization by Example-based Learning" (ECCV 2022)
Apache License 2.0
17 stars 3 forks source link

training on a custom dataset #2

Open surfingnirvana opened 1 year ago

surfingnirvana commented 1 year ago

Is it possible to train my custom dataset with just a small sample size?

winfried-ripken commented 1 year ago

Sure! For parametric style transfer you would only need a pair of style and content images like for classical style transfer. Our local parameter prediction task relies on the code for Pix2PixGAN that accepts all kinds of paired image datasets. Let me know if that answered your question!

surfingnirvana commented 1 year ago

If i try to run the parametric style transfer i get an error:

(base) PS C:\neural\wise> python -m parameter_optimization.parametric_styletransfer --effect xdog --content experiments/source/portrait.png --style experiments/target/watercolor_portrait.jpg Traceback (most recent call last): File "C:\ProgramData\Anaconda3\lib\runpy.py", line 197, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\ProgramData\Anaconda3\lib\runpy.py", line 87, in _run_code exec(code, run_globals) File "C:\neural\wise\parameter_optimization\parametric_styletransfer.py", line 109, in strotss_process(args.content, args.style, effect=effect, preset=preset, cpu=args.cpu) File "C:\neural\wise\parameter_optimization\parametric_styletransfer.py", line 92, in strotss_process result = execute_style_transfer(s, t, resize_dim, device="cpu" if cpu else "cuda:0") NameError: name 'execute_style_transfer' is not defined