I developed a neural network to learn to classify various kinds of noise in an image - salt and pepper, Gaussian, Poisson etc.
I want to validate the model on unseen data. Are there any labelled datasets that I can use?
Example: A labelled dataset of images only affected with Gaussian noise, labelled dataset of images only affected with salt and pepper noise, labelled dataset of images only affected with Poisson noise. It might not be pragmatic to look for labelled datasets where the images are affected by only a particular type of noise, but I am looking for the dominant noise in each set of images to be one of the three.
Some information on what kind of datasets I can use for the testing phase would be much appreciated.