Project funded by a grant of the Ministry of Research, Innovation and Digitization, CCCDI - UEFISCDI, number PN-IV-P6-6.3-SOL-2024-2-0227, within PNCDI IV.
Host institution: University of Bucharest
With the development of adversarial generative networks, generating fake media content has become easily accessible to the general public, especially through the emergence of free software tools. In this project, we aim to develop a web platform for detecting counterfeit media content in image, video and audio files. We will analyze various features for each type of content (image, video, audio), such as artifacts, PRNU signatures, facial expression, speech and lip-sync correspondence, etc., and then develop artificial intelligence (AI)-based systems to detect such features automatically. We will consider developing efficient models that ensure near-real-time operation, as well as compatibility with various file formats and protocols for live streams. The developed systems will be integrated into a suite of web services, which will in turn be used by a user-friendly web application. For proper testing of the AI-based models, we will create 3 test datasets with counterfeit media content, and a web application for generating such content. Finally, we will integrate all developed components (models, services, web applications, databases) into a unified platform.
UNIBUC TEAM
Prof. Radu Tudor Ionescu
Principal Investigator
Assoc. Prof. Marius Popescu
Senior Researcher
Assoc. Prof. Paul Irofti
Senior Researcher
Assoc. Prof. Cristian Rusu
Senior Researcher
Vlad Hondru
PhD Student
Andrei-Iulian Hîji
PhD Student
Florinel-Alin Croitoru
PhD Student
Nicolae-Cătălin Ristea
PhD Student
CREATIVEAST TEAM
Ionuț Dumitran
Co-Principal Investigator
Anca Tache
Software Engineer
Cristina Noaică
Software Engineer
Daniel Ardelean
Software Engineer
Alexandru Canavoiu
Software Engineer
PAPERS
FA Croitoru, V Hondru, RT Ionescu, M Shah
Computer Vision and Image Understanding, 2024
DN Grigore, MI Georgescu, JA Justo, T Johansen, AI Ionescu, RT Ionescu
IEEE/CVF Winter Conference on Applications of Computer Vision, 2025
FA Croitoru, AI Hiji, V Hondru, NC Ristea, P Irofti, M Popescu, C Rusu, RT Ionescu, FS Khan, M Shah
ArXiv, 2024
CODE
Repository for the "Reverse Stable Diffusion: What prompt was used to generate this image?" paper is available here.
Repository for the "Weight Copy and Low-Rank Adaptation for Few-Shot Distillation of Vision Transformers" paper is available here.
Repository for the "Deepfake Media Generation and Detection in the Generative AI Era: A Survey and Outlook" paper is available here.
SCIENTIFIC REPORTS
Technical report for the first stage is available here: Romanian version.