M. NARESH, J RAJU, CH AJAY, D MAHESH, S UDAY KUMAR

Abstract

Using Independent Component Analysis, this work aims to improve Novel Iris Recognition while lowering false rejections. Improved person identification accuracy and decreased frequency of erroneous rejections of linked faces are achieved by the use of Real-Time Face recognition, Mask Detection, and Iris Detection models in this study. The CNN that was utilized to build it was a Python-based one that relies on VGG16. Twenty participants were split evenly between Group 1 and Group 2 for this investigation. With a success rate of 94.266 percent, the proposed Novel Iris Recognition approach outperformed Independent Component Analysis' 93.128 percent. There can be no doubt that the two are completely different. Both methods were statistically equivalent when looking at the major study outcomes for accuracy and loss (p = 0.641, p>0.05). Adding a VGG16 convolutional neural network with 16 layers helped the model perform better. Conventional facial recognition algorithms have difficul