IRIS FaceMatch: A secure face-based identity detector without racial bias using deep learning

This project aims to address the race bias of face recognition technology by developing new face feature sets and building the suggested models based on an equal number of images from varied-race photos using appropriate deep learning algorithms. This research seeks to deliver industry partner IRIS with guidelines for a developed prototype that facilitates the adoption of IRIS’s FaceMatch technology for enhancing police capabilities to find unidentified individuals based on their photos while saving labor force and other police resources. Fixing this bias enables IRIS to serve police at different provinces and extend the usage of the IRIS FaceMatch to other applications, such as searching for friends based on their photos.

Intern: 
Siraj Hamza
Faculty Supervisor: 
El Sayed Mahmoud
Province: 
Ontario
University: 
Partner University: 
Discipline: 
Program: