Deep Learning-Based Automated Signature Verification for Fraud Detection

Abstract

Every individual has a distinctive signature, making it a key element in personal identification and authentication for legal and financial transactions. However, manual signature verification is often labor-intensive and prone to errors, making it a challenge in combating document forgery and falsification.To address this, we have developed an automated online signature verification system utilizing Convolutional Neural Networks (CNNs), a powerful Deep Learning algorithm. The model is trained on over 100 signature samples per user, ensuring high accuracy in detecting forged signatures through extensive testing.The system is built using the Flask framework, combining HTML, CSS, and JavaScript for a user-friendly interface, while the backend integrates Python and a MySQL database for efficient data management and secure authentication. This approach provides a fast, accurate, and scalable solution for online signature verification, enhancing security in digital transactions.