Multi-Task Deep Learning for Multimodal Biometric Recognition
Said Si Kaddour, Larbi Boubchir, Boubaker Daâchi · 2022
Automatic person recognition systems are mainly based on biometrics traits to authenticate or identify people in different real-life cases. In order to get reliable results, all the web giants prioritize investing huge sums of money in these studies. In 2019, the National Institute of Standards and Technology (NIST) published outstanding achievements in facial recognition technology based on facial biometrics. One recent study estimates that the facial recognition market will be worth 8 billion euros by 2024, with double-digit annual growth rates. The success of biometric recognition has been driven by the advent of neural networks such as deep learning architectures. These learning methods quickly become popular and have offered an interesting solution for realizing intelligent biometric recognition systems. Our work aims to study the use of convolutional neural networks for biometrics recognition based on various biometric traits (such as face, palmprint, and palm vein), and also to design a multi-task learning model for multimodal biometric recognition,