Spot the difference (IMAGE)

University of Tokyo

Spot the difference

Caption

An example of some deepfake photos were made using different manipulation methods (DF, F2F, FS and NT). A deepfake detector was then trained using an established data set of sample deepfakes (FF++), while a duplicate detector was trained using the researchers’ self-blended images (SBIs). The two detectors were given the above deepfake photos. The columns of false color images show the difference between training using existing data sets and training using SBIs.

Credit

©2022 Yamasaki and Shiohara

Usage Restrictions

Credit ©2022 Yamasaki and Shiohara

License

Original content

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.