The biometric research in SAFE investigates a broad range of biological and behavioural biometrics, including 2D and 3D face recognition, fingerprint recognition, finger-vein recognition, voice biometrics, keystroke recognition, gesture recognition and mouse dynamics. SAFE builds on this expertise to develop trustworthy identity verification that can resist digital manipulation while remaining usable, explainable and fair.

Focus themes

• Presentation attack detection (PAD): methods that recognise attempts to deceive biometric systems using presentation attack instruments, including previously unseen attacks.
• Injection attack detection (IAD): protecting image, video and sensor capture pipelines against injected media and emerging manipulation such as deepfakes.
• Generative and synthetic face generation: using generative models to create realistic and diverse synthetic faces for algorithm development, privacy-aware experimentation and robustness testing.
• Novel and multimodal biometrics: combining biological and behavioural signals across smartphones, kiosks, digital banking, electronic identity and other high-value services.
• Fairness and explainability: improving identity verification across demographic groups and making automated decisions easier to understand and evaluate.

Synthetic biometric data generation

SAFE develops methods for generating synthetic biometric data, with particular emphasis on face data, to support privacy-aware research, controlled testing and evaluation where representative real-world data are limited or sensitive. The work examines data quality, diversity, realism and the responsible use of generated samples for training and testing biometric systems.