1
$\begingroup$

I have input dataset in the form of images and output data is also an images insteade of being labeled data. So it looks neither classification problem nor regression problem. Input and output iamges may have some correlation between them and I want my model to learn that correlation. I am struggling to find the proper way of implementing this. Can anyone help with that?

$\endgroup$
1

1 Answer 1

0
$\begingroup$

Input and output iamges may have some correlation between them and I want my model to learn that correlation

In order to calculate "correlation" (more likely some kind of similarity) you must have the two images as input, and the correlation/similarity score is the output.

Maybe what you mean is that you want to predict the most similar images to a particular input image. This would be a ranking task, it's usually done by comparing the input image to a predefined set of images. The system predicts a similarity score for every image and returns the top similar ones.

$\endgroup$
2
  • $\begingroup$ Actually I dont have two image as an input, but if it is possible to feed two image as an input then I think it would serve the purpose. I will elaborate my question, let say X is image of some 4 limb animal and y i also an image of another 4 limb animal, then I want my model to learn that kind of similarity taht both has 4 limb. $\endgroup$ Jul 2, 2021 at 19:14
  • 1
    $\begingroup$ @ChiragPalan it sounds like you have a set of images as input and you want to find similarities among these images, possibly group them by similarity. In this case this would be a clustering problem. $\endgroup$
    – Erwan
    Jul 2, 2021 at 22:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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