Lip Print Identification Using Deep Learning

Abstract

The idea behind biometric identification is that every person has distinct qualities. While iris, retina, fingerprint, and face recognition are common traditional ways, an alternative Lip-based identification is a method that uses the permanent patterns on human lips as a biometric measurement. It takes a lot of work to physically collect lip prints. By taking higher-quality pictures rather than relying on lip print detection, we can accomplish this more quickly and efficiently using convolutional neural networks & computer vision. The purpose of this research identifies people only by their lip prints. Our suggested approach makes use of the convolutional neural network (CNN) capacity to recognize and extract distinctive characteristics from lip prints. We use data augmentation and transfer learning approaches to overcome the issues of limited data availability and variability in lip print images. Several deep learning models were employed here. With Resnet50, I achieved 97% accuracy. It was discovered that employing VGG 19 and achieving 93% accuracy as well as using Inception Net and achieving 98% accuracy..