Building a fully functional iris recognition system

The research in my lab is primarily focused on human biometrics and security, especially on iris recognition and methods of detecting unknown presentation attacks. But we conduct various projects in computer vision, machine learning and neural networks areas, with their non-obvious intersections with psychology, medical sciences, and art. 

The goal of this particular project is to build a fully functional, open-source, and open-hardware iris recognition demo with presentation attack detection capabilities. The system will be built on a Raspberry Pi (preferably) or an NVIDIA Jetson (if we decide that it’s needed) platform. We will utilize open-source tools developed by the AI Trust and Reliability (AITAR) and Computer Vision Research (CVRL) Labs at Notre Dame, including models for iris image segmentation, encoding, and deep learning-based detection of fake eyes.

A good starting point is this IEEE IJCB paper: https://arxiv.org/abs/2008.08220. Additionally, this GitHub repository contains the open-source solutions available for implementation: https://github.com/CVRL/OpenSourceIrisRecognition

Students working on this project will learn how iris recognition works and how to deploy deep learning models on low-resource platforms such as the Raspberry Pi. If successful, this project will yield a fully functional demo to showcase biometrics-related research conducted at Notre Dame (with full credit given to the creator in all presentations).

If you are interested in this project, please do whatever is required here in the system, but (most importantly) please schedule a meeting with me (just send me an email at aczajka@nd.edu and we will find a good time to meet). I cannot make admission decisions by only looking at a CV, and I prefer to first meet with students to discuss their interests, expectations, and backgrounds. This gives us a much better chance of putting you on a project that you will love and that will allow you to learn the things you want.