Comparison among different CNN Architectures for Signature Forgery Detection using Siamese Neural Network

Abstract

Signature is the most common way to indicate knowledge and acceptance of a document. As many documents and contracts are now starting to use paperless electronic formats, the term "signature" has been substantially broadened. Whichever form it takes, the key importance of the signature is identity authentication for managing security. Signatures being one of the most widely used methods for the same, play a crucial role in financial, legal, and social aspects of one's life. Thus, Signature forgery, that is falsely copying another individual's signature is an issue of utmost concern. The chances of two or more signatures made by the same individual being identical are minimal, thus making signature forgery detection an arduous task. Our paper aims to apply the state-of-the-art methodology, Siamese Neural Networks, on the chosen data set, draw meaningful insights and perform a comparative analysis between some variants of these neural networks to identify and authenticate handwritten signatures.