Handwritten Signature Recognition using MobileCNN
Kudupudi Durga Harika, P .M. Dhanalakshmi, Singam Sharan Kumar Reddy, Shaik Mahamad Shakeer, M V SreeVishnu Thanmayi, V Haripriya · 2024
Classifying handwritten signatures becomes important when it comes to document verification. In order to maximize the precision and effectiveness of the signature examination process, this research presents a robust method for classifying handwritten signatures that makes use of Convolutional Neural Networks, and the light-weight Mobile Net architecture. Because of the inherent differences and quirks in each person's handwriting style, classifying signatures is a difficult operation that requires a method that can recognize and learn these subtle differences. By integrating a CNN, the suggested model MobileCNN successfully extracts hierarchical characteristics from the signature photos. The Mobile Net design also guarantees the model's portability and adaptability to a range of devices, including ones with little processing power. Additionally, our method makes use of transfer learning and data augmentation techniques to improve the model's performance and generalization skills on unseen data. According to experimental findings, the suggested signature classification system exhibits impressive accuracy and low latency, which qualifies it as a viable option for real-time use in the domains of identity authentication and secure document verification.