Handwritten Signature Recognition using Deep Learning
Basmala Mustafa, Radwa Taha, Omar M. Fahmy, Shereen Moataz Afifi · 2023
Handwritten signature recognition plays a crucial role in verifying document authenticity and preventing fraudulent activities. That’s why this paper focuses on the development of a deep learning-based system for recognizing handwritten signatures. The main objectives include creating a diverse dataset of signatures, implementing a deep learning architecture for accurate signature recognition, and evaluating the system’s performance using various metrics.The VGG16 architecture was chosen due to its effectiveness compared to other methods, and it served as the framework for further enhancements. The results demonstrate the model’s outstanding accuracy. Specifically, the proposed model, trained on the merged dataset, achieved remarkable performance with a training accuracy of 99.78% and a validation accuracy of 99.75%. During testing, the model exhibited an impressive accuracy of 98.96%, confirming its effectiveness in identifying genuine signatures. Furthermore, the model trained on the collected dataset had shown an accuracy of 98.9% which ensures an efficient handwritten signature recognition.