Cloud-based Signature Validation Using CNN Inception-ResNet Architecture
Chieri Ishikawa, Jeff Allen U. Marasigan, Meo Vincent C. Caya
Abstract
Signature validation is a critical component of biometric security systems, especially those used in high-risk sectors like finance or legal. To reduce the cases of document falsification and forgery, a method of conducting digital signature verification has been devised. The developed system makes use of the Inception-ResNet neural network model to create a cloud-based system that conducts signature validation via image processing techniques. Upon testing, the developed system obtained an overall accuracy percentage of 85%, as well as False Acceptance Rate (FAR) and False Rejection Rate (FRR) of 20.83% and 6.25%, respectively. Upon conducting a comparison test with similar existing methodologies, it was concluded by the proponents that the developed system possessed better accuracy as well as lower error rates than previous research regarding signature validation, making the prototype a viable alternative to current verification methods. It was also recommended by the proponents that in further studies on the topic, more standardized references should be trained in the neural network model to further improve system accuracy.