Denoising an image using Noise2SelfCNN restoration API

One can use the following lines to denoise a single image with default options using our Object-Oriented denoising API.

from aydin.restoration.denoise.noise2selfcnn import Noise2SelfCNN

n2s = Noise2SelfCNN()
n2s.train(noisy_image)
denoised_image = n2s.denoise(noisy_image)

It is also easy to pass specific transforms to use before and/or after denoising. One can do the the following:

from aydin.restoration.denoise.noise2selfcnn import Noise2SelfCNN

transforms = [
     {"class": RangeTransform, "kwargs": {}},
     {"class": PaddingTransform, "kwargs": {}},
 ]
n2s = Noise2SelfCNN(it_transforms=transforms)
n2s.train(noisy_image)
denoised_image = n2s.denoise(noisy_image)

One can also use the following lines to denoise a single image with default options using our procedural denoising endpoint.

from aydin.restoration.denoise.noise2selfcnn import noise2self_cnn

denoised_image = noise2self_cnn(noisy_image)