Computer Vision · Biometrics · Media Forensics

Luis F. Gomez

Ph.D. researcher working on computer vision, face biometrics, media forensics, and AI-based video analysis.

My work explores how visual representations can be used to analyze identity-related information in real and synthetic media. I am especially interested in the role of foundation models, temporal features, and biometric cues for forensic and security-oriented applications.

I also work on video-based biomechanical analysis, focusing on markerless pose estimation and movement assessment from running videos.

Research Interests

View all publications

Facial Expression Biometrics

Facial behavior analysis for healthcare, e-learning, and biometric applications

This research line explores facial expressions and facial action units as biometric and behavioral cues for human-centered applications. The work focuses on extracting and analyzing facial dynamics to support tasks such as health assessment, attention estimation, and identity-related analysis.

A particular focus is placed on the use of facial behavior for Parkinson detection, e-learning scenarios, and multimodal face biometrics, connecting computer vision methods with real-world healthcare and educational applications.

Deepfake Biometric Security

Robustness of face biometric systems against digital attacks

This research focuses on the challenges posed by digital attacks, especially deepfakes, in face biometric systems.

The objective is to analyze vulnerabilities in face verification scenarios and propose methods to improve the robustness and security of biometric systems against manipulated or synthetic visual content.

Foundation Models for Media Forensics

Using visual representations for biometric and forensic analysis

This research line explores visual features extracted from foundation models such as DINOv2 and CLIP for forensic and biometric applications.

The focus is on understanding whether these representations capture identity-related, appearance-based, or dynamic information, and how they can be combined with task-specific models.

Avatar Fingerprinting

Identity traces in photorealistic talking-head avatar videos

This research line studies whether photorealistic avatar generation and facial animation pipelines preserve, distort, or introduce identity-related biometric traces.

The goal is to understand how synthetic talking-head videos affect biometric verification, identity analysis, and media forensic applications.