Signature verification is crucial in many real-world applications, such as banking and legal documents. However, conventional approaches that have been followed for many years are susceptible to forgeries. The accuracy of these models should be sufficiently high since they are used by banks and courts, where signatures act as prominent evidence in validating the identity of an individual. This paper proposes a novel ensemble learning approach to signature authentication using two robust deep learning models. Ensemble learning leverages the complementary strengths of CNNs and Siamese networks to achieve higher accuracy and robustness in signature verification. It seamlessly accommodates images, reducing the likelihood of false positives and false negatives. This growth in the development of signature verification systems could mitigate the potential for forgery and the involvement of any third-party person who judges the validity of the signatures.