Deepfakes: between reality and imagination
The enormous advances in Artificial Intelligence have made it possible to develop advanced technologies for image and video analysis.
To date, we have intelligent systems capable of recognizing objects, analyzing the movement and interactions among people in a given setting, recognizing human emotions through face analysis, and much more.
But the most surprising developments in recent years have been in the creation of video content. Thanks to modern AI technologies it is now possible to create artificial but extremely realistic videos, called deepfakes, which portray people or objects that do not actually exist or that move in virtual contexts.
In recent years our research activity has focused on this new frontier of computer vision and we have developed new deep learning models that can be used to create artificial videos by animating static images or manipulating real videos. In particular, we have developed an innovative technology, the First Order Motion Model (FOMM), capable of generating deepfakes in real time.
- Paolo Carta, Director of Collegio Clesio
Elisa Ricci is associate professor at the Department of Information Engineering and Computer Science (DISI) of the University of Trento and head of the Deep Visual Learning research group of the Bruno Kessler Foundation (FBK). Her research interests are mainly in the area of computer vision, robotics, machine learning and human behaviour analysis.