Dr. Mahdi Ghafourian

I develop a real-time head pose estimation system for InfinitySet, enabling AR/VR scene interaction via TV presenters’ head movements. My method works well in low-data settings, relevant for costly data scenarios like autonomous driving. I’ve gained expertise in real-time pipelines, data-efficient learning, and system integration. This work is part of the EMERALD project (Horizon Europe 101119800) in the IMVA group, focusing on data-efficient pose estimation.

Funded by a Marie Skłodowska-Curie fellowship (PriMa), I completed a Cum Laude PhD in biometric recognition, privacy-preserving AI, and adversarial machine learning. I focused on secure face biometrics and privacy tech, gaining experience in deep learning (PyTorch, TensorFlow), end-to-end pipelines, high-dimensional biometric data handling, data augmentation, hyperparameter tuning, and large-scale ML experiment design, with publications in top-tier venues.

Biometric Protection

Cryptography, Trust, Security protocol

Web programming, Robotic