How to implement Noise2Noise using ImageTranslatorFGR
It is quite easy to train Noise2Noise model with image translators provided in Aydin API. You can see an example below using ImageTranslatorFGR:
from aydin.it.fgr import ImageTranslatorFGR
it = ImageTranslatorFGR()
for noisy_image1, noisy_image2 in noisy_pairs:
denoised = it.translate(noisy_image)
Noise2Noise is a great method to train a model from pairs of images sharing the same information signal with different noise instances. You can find more information about how to prepare noisy image pairs on the original paper. More information about the paper is available here and the code that is published with paper can be found on github.