MuDDi - Multi Modal Deepfake Detection

Unmasking Deception, Preserving Trust.

What is MuDDi?
#

MuDDi (Multi-modal Deepfake Detection) is a digital forensic tool designed for analyzing and detecting manipulated content in videos and images. Using state-of-the-art AI technology, MuDDi identifies deepfakes and provides reliable assessments of media authenticity.

Whether in digital forensics, fact-checking, or law enforcement, MuDDi helps uncover manipulated content and safeguard the integrity of digital media.

How Does MuDDi Work?
#

MuDDi integrates advanced AI methodologies for precise deepfake detection:

🔍 Videoplethysmography (VPG): Analyzing micro-movements in skin regions to detect missing or unnatural pulse waves in deepfake videos.
📸 Image Forensics & Watermark Analysis: Identifying manipulations through high-resolution image feature analysis and invisible watermark detection.
🧠 AI-Powered Expert Evaluation (LLMs): Large Language Models assess media authenticity based on stylistic and contextual analysis.

Where is MuDDi Used?
#

MuDDi is specifically designed for professional use in digital forensics and supports:

Fact-checkers & Media Organizations – for verifying videos and images.
Security & Law Enforcement Agencies – for digital evidence collection and authentication.
Platform Providers & Tech Companies – to combat misinformation and media manipulation.

MuDDi’s technology is set to be deployed in real-world applications, including CORRECTIV’s Faktenforum. This participatory platform allows citizens and experts to work together to identify manipulated content and disinformation.

Award-Winning Technology: SPRIND Innovation Competition
#

MuDDi is one of the winners of the SPRIND Funke Innovation Challenge “Deepfake Detection & Prevention” – a testament to the technological relevance and innovation behind our approach.

Applied AI Research for Digital Democracy
#

MuDDi is developed at the FZI Research Center for Information Technology, a leading independent research institution for applied computer science. As a non-profit institute, FZI combines cutting-edge research with practical applications, supporting businesses, government agencies, and civil society in developing innovative technologies.

The project is led by the House of Participation (HoP) at FZI – an interdisciplinary research hub for digital democracy. HoP develops and investigates technological solutions to strengthen democratic processes, combat disinformation and deepfakes, and support a future-oriented digital public sphere and governance.

Our Team
#

MuDDi is an interdisciplinary research project at the FZI Research Center for Information Technology, bringing together expertise in AI, media forensics, and digital democracy.