Multimedia Forensics and Security

The Multimedia Forensics and Security (MFS) is a research group dedicated to advancing the state of the art in Artificial Intelligence for Digital Forensics, Multimedia Analysis, and Security. MFS focuses on the development of methodologies for the acquisition, analysis, interpretation, and validation of digital evidence, with particular attention to multimedia data and complex forensic scenarios.

The mission of MFS is to design, develop, and evaluate robust and trustworthy solutions for addressing emerging challenges in digital investigations, evidence integrity, and information authenticity. The group operates at the intersection of computer vision, machine learning, and forensic science, integrating technological innovation with investigative and legal requirements in high-stakes environments.

MFS conducts research on the analysis of visual, audio, and multimodal data, covering the entire digital forensic pipeline, from data acquisition and preservation to analysis, interpretation, and presentation of evidence. Core research activities include the detection and analysis of manipulated content, authenticity verification, and source attribution, as well as the study of synthetic media such as deepfakes and generative models.

Beyond multimedia analysis, MFS addresses digital forensics in a broader sense, including methodologies for evidence collection, chain of custody, and forensic workflows aligned with legal and investigative standards. The group explores interdisciplinary domains such as virtual reality forensics, forensic firearm ballistics, and handwritten document analysis, dealing with both digital and hybrid forms of evidence.

Particular attention is given to fraud detection, anti-forensic techniques, and the development of countermeasures aimed at ensuring the reliability, robustness, and admissibility of digital evidence in judicial contexts. In this framework, MFS investigates adversarial forensics, focusing on the analysis of attacks against forensic algorithms, the robustness of detection systems under adversarial conditions, and the design of resilient methodologies capable of operating in hostile environments. MFS also explores the implications of emerging technologies, including generative AI, in the context of security, misinformation, and forensic analysis.

The group further explores advanced machine learning paradigms, focusing on robustness, generalization, and explainability of AI systems in forensic applications. Research includes domain adaptation, cross-modal analysis, and the integration of AI into forensic workflows, as well as the study of adversarial machine learning techniques and their impact on forensic reliability, with the goal of supporting investigators in complex and high-risk decision-making processes.

Key research topics include:

• Deepfake detection, localization, and attribution;
• Multimedia authenticity and integrity verification;
• Generative model analysis and forensic traceability;
• Image, video, and multimodal forensics;
• Virtual Reality (VR) forensics and immersive environments;
• Forensic firearm ballistics analysis;
• Handwritten document analysis and authorship attribution;
• Fraud detection and anti-forensic techniques;
• Explainable and trustworthy AI for forensic applications;
• Domain adaptation and generalization in forensic models;
• Cross-modal and multimodal forensic analysis;
• Digital evidence interpretation and forensic workflows.
• Adversarial forensics and resilience of forensic systems under adversarial conditions;

MFS is part of the Image Processing LABoratory (IPLAB) and actively collaborates with academic institutions, industry partners, and law enforcement agencies. Through these collaborations, MFS contributes to the development of secure, reliable, and human-centric forensic technologies, while supporting the training of students and researchers in the field of AI-driven digital forensics.