News

2018/09/20

Assoc. Prof. Vincent Nozick (NII past visiting researcher), Assoc. Prof. Junichi Yamagishi, Prof. Isao Echizen and others in a research team develop new deep learning technology for detecting fake videos

Japan's National Institute of Informatics (NII; Chiyoda-ku, Tokyo; Dr. Masaru Kitsuregawa, Director General) announces that a research team consisting of past visiting researcher Associate Professor Vincent Nozick (Université Paris-Est Marne-la-Vallée, France), Associate Professor Junichi Yamagishi of Digital Content and Media Sciences Research Division, Professor Isao Echizen of Information and Society Research Division, and past research student Darius Afchar (master course student, École Nationale des Ponts et Chaussées, France) has developed new deep learning technology for automatically detecting fake videos skillfully created by techniques that automatically alter faces in videos.

MesoNet: a Compact Facial Video Forgery Detection Network
https://arxiv.org/abs/1809.00888

This technology focuses on techniques such as Deepfake and Face2Face used for generating realistic fake videos. Evaluation experiments revealed detection accuracy greater than 98% for Deepfake and 95% for Face2Face.

The results of this research are scheduled to be presented at the 2018 IEEE International Workshop on Information Forensics and Security (WIFS 2018) to be held from December 11, 2018.

This research was supported by Japan Society for the Promotion of Science (JSPS) Grants-in-Aid JP16H06302, JP18H04120, and JP18KT0051.