Open Media Forensics Challenge Evaluation (OpenMFC) is an open evaluation series organized by the National Institute of Standards and Technology (NIST) to assess and measure the capability of media forensic algorithms and systems.
Due to remarkable research and interest growth in Media Forensics in recent years, the NIST team initiated a new leaderboard evaluation platform, OpenMFC (https://mfc.nist.gov/) in 2020. The objective of OpenMFC is to help advance the-state-of-the-art of media forensics technologies, to open a challenge evaluation series to the public for all researchers and make participation easy and convenient, to support the research with the evaluation benchmark datasets, the evaluation infrastructure, and the communication forum, to provide the community with rich resources to aid in the development of the technologies, and thus to promote media forensics research worldwide.
OpenMFC focuses on the performance measurements of automated imagery (image and video) manipulation detection and localization technologies that determine if the media is manipulated, and if so, the region and type of manipulation in a given media. OpenMFC also supports Generative Adversarial Network (GAN) manipulation detection tasks.
OpenMFC is based on the experiences we collected in the Media Forensic Challenge (MFC) evaluation project for the Defense Advanced Research Projects Agency (DARPA) Media Forensics (MediFor) program (http://www.darpa.mil/program/media-forensics). MFC (2017-2020) is an evaluation series developed under the MediFor program to focus on the performance evaluation of DARPA MediFor research teams. Before the evaluation, NIST released the Nimble Challenge’16 (NC2016) dataset as the program kickoff dataset, (where NC is the former name of MFC). The first evaluation of the DARPA MediFor MFC series took place in the summer of 2017 (NC2017); The second evaluation took place in the summer of 2018 (MFC2018); The third evaluation took place in the summer of 2019 (MFC2019); and the fourth evaluation took place in the summer of 2020. Latest MFC evaluation also supports other media forensic related evaluation tasks such as constructing a phylogeny graph describing the manipulation history of an image; verifying the media sensor identification, etc.
If you would like to know more about OpenMFC, to obtain NIST-released datasets, or to participate in the evaluation, please visit our website: https://mfc.nist.gov. OpenMFC is releasing both NC16 kickoff dataset and NC17 evaluation dataset to the public upon request via signing up on the website. Perspective OpenMFC participants can sign up on the website and complete the license agreements. If you have any questions or comments, please send an email to the contact below.