“Forensic Similarity for Digital Images,” an article by MISL member Owen Mayer and MISL PI Dr. Matthew Stamm was recently selected to be published in the IEEE Transactions on Information Forensics and Security, the #1 academic journal for Information Security. The article is to appear in print in December 2019.
The article describes a new deep learning technique to reliably determine whether two small image regions contain the same or different forensic traces, which are used to expose manipulated multimedia, falsified images, and fake news.
Visit the project website
for more information and interactive demos. Code can be found on the git repository