Deepfake refers to using artificial intelligence (AI) and machine learning techniques to create compelling and realistic media content, such as videos, images, or recordings, that appear real but are fake. The most common form of deepfake involves using deep neural networks to replace or superimpose faces in existing videos or images on top of other people’s faces. While this technology can be used for various benign purposes, such as filmmaking or online education, it can also be used maliciously to spread misinformation by creating fake videos or images. Based on the classic deepfake generation process, this paper explores the Inconsistency between inner and outer faces in fake content to find synthetic defects and proposes a general deepfake detection algorithm. Experimental results show that our proposed method has certain advantages, especially regarding cross-method detection performance.

Generalized Deepfake Detection Algorithm Based on Inconsistency Between Inner and Outer Faces

Concas, Sara;Orru', Giulia;Marcialis, Gian Luca;Roli, Fabio
2024-01-01

Abstract

Deepfake refers to using artificial intelligence (AI) and machine learning techniques to create compelling and realistic media content, such as videos, images, or recordings, that appear real but are fake. The most common form of deepfake involves using deep neural networks to replace or superimpose faces in existing videos or images on top of other people’s faces. While this technology can be used for various benign purposes, such as filmmaking or online education, it can also be used maliciously to spread misinformation by creating fake videos or images. Based on the classic deepfake generation process, this paper explores the Inconsistency between inner and outer faces in fake content to find synthetic defects and proposes a general deepfake detection algorithm. Experimental results show that our proposed method has certain advantages, especially regarding cross-method detection performance.
2024
978-3-031-51022-9
978-3-031-51023-6
Deepfake detection; Generalization; Manipulations
File in questo prodotto:
File Dimensione Formato  
ICIAP2023deepfakegeneraliz (1).pdf

embargo fino al 24/01/2025

Tipologia: versione post-print
Dimensione 2.89 MB
Formato Adobe PDF
2.89 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/389063
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact