NIST Open Media Forensics Challenge (OpenMFC)

 NIST Open Media Forensics Challenge (OpenMFC)

A well-defined evaluation is necessary to rapidly measure accuracy and robustness of systems over diverse datasets collected under various environments and to advance technologies that meet real-world scenarios.

Due to recent remarkable growth of research interest in media forensics, the National Institute of Standards and Technology (NIST) launched a web-based leaderboard challenge platform, Open Media Forensics Challenge (OpenMFC) in 2020. The OpenMFC is an open evaluation to promote development of AI-based systems that can automatically detect and locate manipulations and deep-fakes in images and videos worldwide. The evaluation includes three major challenges; 1) Image Manipulation Detection & Localization, 2) Video Manipulation Detection, and 3) Deep Fakes Detection & Localization (GAN: Generative Adversarial Network). 

NIST has a series of large datasets that were collected for supporting the Defense Advanced Research Projects Agency (DARPA) Media Forensics (MediFor) program over 4 years (2017-2020) and a subset of the datasets was released to over 300+ researchers, 150+ organizations, and 26+ countries worldwide. Using a structured evaluation infrastructure, the OpenMFC supports each challenge using designed datasets, conducts experiments and data analyses to understand system behavior, and provides research direction and guidance for system improvements.

The OpenMFC aims to engage the larger research community using the web-based Leaderboard to serve the participants worldwide. For details, please visit the OpenMFC website at https://mfc.nist.gov/.