about the results of CLTS-GAN

[rhyhck]

Hello, author. When I was trying to reproduce CLST-GAN, I used the dataset and model weights from the link you provided. However, the results I obtained are significantly different from the results you showed. Could you please help me understand the reason for this? Here is the link you provided: https://zenodo.org/records/7036198.

[sanadeem]

Thank you for reaching out. Unfortunately, the student who worked on this left a while back. We will be revisiting this repo when a new student joins. Leave a comment on what is different. This is a stochastic model and you will see different results every time you run the model with different noise. I see changes in specular reflection/texture between the images you posted which means the model is working for the specular/texture noise vector. Our observation was that the changes will be more pronounced if you start from the virtual colonoscopy gray scale image rather than the OC image.