• The aim of this course is to introduce students to the fundamentals of biometrics, that is to say, the identification of an individual from physical or biological characteristics.
• From biometric data acquisition and biometric characteristics extraction, learning of recognition and authentication systems, up to fake, fraud, and their detection.
• A particular attention will be paid to critical concepts for biometrics such as reliability and bias, and will make students aware of its legal and judicial aspects.
• This course is part of the Master of Engineering in Intelligent Systems from Sorbonne Université.
• Students are strongly encouraged to have taken classes in digital sound and image processing (filtering, mathematical morphology, short-term Fourier transform, etc.), machine learning and neural networks, and programming in Python language.
• Knowledge of sensors and instrumentation is a plus.
• A brief reminder of the fundamentals will be given at the beginning of the course if necessary.
C1 (2h) – Introduction to biometrics (civil and police systems, performance measures). Iris recognition:
segmentation, characteristic points extraction, comparison of iris-codes.
C2 (2h) – Fingerprints: discriminating characteristics, extraction and comparison, recognition systems.
C3 (2h) – Facial recognition: characteristic points extraction, eigen-faces (PCA/LDA), face detection, deep learning for face comparison.
C4 (2h) – Speaker authentication: specificities and approaches.
C5 (2h) – Video analysis: objects detection and tracking in videos, applications to video surveillance.
C6 (2h) – Biometric sensors: cameras, lighting, fingerprint sensors (prism optics, silicon, TFT).
C7 (2h) – Fraud and deep fake: neural architectures for the generation and falsification of biometric data (GAN, AttGAN, Fader Network). Applications to face generation and images, voice, and videos manipulation.
C8 (2h) – Biometric fraud detection: typology of fraud, detection methods (faces, fingerprints).
C9 (2h) – Biases and ethics in biometrics: potential biases, societal risks and their regulation. Example of judicial expertise in voice identification and protection of personal data.
Lectures notes & videos.
Jupyter notebooks: illustrations, exercises.
Access to the GPU server (in collaboration with the Sorbonne Center for Artificial Intelligence) Room 226, Esclangon building.
Resources and information on Moodle.