Building a Face Recognition Demonstrator
Group: Responsible AI for Biometrics
Face recognition is one of the most common biometric technologies used in applications like home security, phone locks, surveillance, and border security. Due to its user-friendliness and minimal human interaction, face recognition systems have replaced many conventional identity authentication methods, including passwords and cards. Face recognition involves detecting faces in images or videos and identifying identities by matching a probe face against a database of enrolled faces. Building a secure end-to-end face recognition system requires the deployment of various modules such as face detection and alignment, face features extraction, and face matching. There are also other modules such as face image quality and face presentation attack detection which complement the face recognition pipeline by improving its performance and enhancing the system's security.
The main goal of this project is to create a demonstrator that showcases real-time face recognition technology through deploying and integrating the different modules of an end-to-end face recognition system. This includes face images pre-processing (detection and alignment), feature extraction, identity matching, quality assessment and presentation attack detection.
Resources:
For further information about the project tasks and requirements, please check out the Project Group presentation (Presentation available here).
Also, these works for reference:
-
Qmag-face: Simple and accurate quality-aware face recognition.
-
SER-FIQ: Unsupervised estimation of face image quality based on stochastic embedding robustness.
-
Joint face detection and alignment using multitask cascaded convolutional networks.