0
$\begingroup$

I have to do a project on signature verification. My goal is having a program with two images as inupt (one is a genuine signature of a selected subject and the other one is the one that I want to test). I have a dataset to train the algorithm with 3000 signatures: half of them are genuine and the other ones are forgery, and they are taken from 150 subjects.

My doubt regards the confrontation between the two images I use as input (the genuine one and the one I want to test). How can I deal with it?

$\endgroup$
1
$\begingroup$

I think a Siamese Network is the solution you are looking for. Arxiv has a paper which precisely addresses your problem.

$\endgroup$

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.