Main Article Content
Spread of Covid 2019 pandemic has shattered the globe, putting everybody in panic stage. In view of its long-term impact on daily life, the need to wear face masks and maintain social distances necessitates. In this current satiation everyone need a contactless biometric communication system for all future authentication schemes. One of the alternatives is the use of Contactless Authentication System biometrics because of its non-contact alike normal biometric finger prints and is capable of recognising even people wearing face masks. In this work, a novel hand gesture-based sign digits scheme is proposed. This work presents the design and implementation of a deep learning system that can be verified without contact by the user using 'authentication code.' The 'authentication code' is a 'n' numeric code, and the digits are hand movements of the sign language digits. We propose a memory-efficient deep learning convolution neural network model to identify and classifying the hand movements of the sign language digits and also to extract the function by combing the two BEMD and SIFT algorithm techniques. The model is deployed in the Raspberry pi 4 Model B edge computing system to act as an edge device for user verification. The model achieves a classification accuracy of 98.47 percent for the publicly accessible sign language digits dataset.