search

liuqk3 / PUT

Paper 'Transformer based Pluralistic Image Completion with Reduced Information Loss' in TPAMI 2024 and 'Reduce Information Loss in Transformers for Pluralistic Image Inpainting' in CVPR2022
MIT License
173 stars 15 forks source link

Mask incorrect when running inference inpainting #11

Closed djqualia closed 1 year ago

djqualia commented 1 year ago

Thanks for sharing this code!

I'm running the inference script (inference_complet_sample_in_feature_for_evaluation) and everything seems to work as expected except I have noticed strange handling of the masks. For example, in one case the input mask is:

mask_binary

...but in the inference output the mask is:

8fbefb29-0663-432c-a62c-059ec6f7a5b3

Notice the difference at the edges...

Is this a bug or am I doing something incorrectly?

djqualia commented 1 year ago

My bad, I think this is from bad mask inputs (and converting them to monochrome with .convert('1') fixed the problem)