0
$\begingroup$

Suppose you have a binary outcome variable and have some training data (10,000 images in jpg format). Also you have a test set of say 11,000 images. If we want to train a classification model and want to improve the image quality (denoise the images), should we do it for every image in the training set? Likewise, should we do it for very image in the test set? Or should the images in training set randomly be chosen to be denoised (likewise in the test set)?

$\endgroup$
2
  • $\begingroup$ To me, it's not clear exactly what the question is. You have a binary classifier, but you also want to denoise the images. I don't see how these 2 tasks are related. Can you clarify this point? Moreover, it's not clear to me why you're wondering about denoising training or test images and which ones to denoise. If you want to denoise all images, then denoise all images, no? I am probably misunderstanding something here. $\endgroup$ – nbro Feb 4 at 11:32
  • $\begingroup$ @nbroL Suppose you have a training set with 10,000 images, denoise 0.6 of all of them. If you have 5,000 images, how many should you denoise? $\endgroup$ – convguycon Feb 5 at 19:17
0
$\begingroup$

You can denoise a certain fraction of images (preferably 0.25-0.3) randomly for each epoch. Adding Gaussian noise to images gives better results.Refer this : Link

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.