Abstract
The detection of manipulated videos represents a highly relevant problem in multimedia forensics, which has been widely investigated in the last years. However, a common trait of published studies is the fact that the forensic analysis is typically applied on data prior to their potential dissemination over the web. This work addresses the challenging scenario where manipulated videos are first shared through social media platforms and then are subject to the forensic analysis. In this context, a large scale performance evaluation has been carried out involving general purpose deep networks and state-of-the-art manipulated data, and studying different effects. Results confirm that a performance drop is observed in every case when unseen shared data are tested by networks trained on non-shared data; however, fine-tuning operations can mitigate this problem. Also, we show that the output of differently trained networks can carry useful forensic information for the identification of the specific technique used for visual manipulation, both for shared and non-shared data.
Highlights
Latest advancements in artificial photo-realistic generation enabled new outstanding possibilities for media data manipulations
FaceForensics++ consists of 1000 original videos, each of them manipulated through 5 different manipulation techniques Deepfake (DF), Face2Face (F2F), FaceSwap (FS), NeuralTextures (NT) and FaceShifter (FSH) techniques
In this work we have addressed the challenging scenario where forensics analysis is applied to manipulated videos shared through social media platforms
Summary
Latest advancements in artificial photo-realistic generation enabled new outstanding possibilities for media data manipulations. While several methodologies and datasets have been published during the last years, one rather unexplored aspect is the generalization capability of those deep descriptors in situations where data are shared through social platforms [3,4] This is a known and ever emerging problem in multimedia forensics [5], given the pervasive role of popular social media platforms in the dissemination and exchange of visual content on a daily basis. In this regard, this work presents the results of an extensive detection analysis which goes beyond controlled laboratory conditions, typically adopted in previous works [1], and deals with a scenario where data are analyzed as direct outputs of manipulation algorithms and after upload/download operations through a popular sharing service.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have