Enhanced Deep Learning DeepFake Detection Integrating Handcrafted Features
By: Alejandro Hinke-Navarro , Mario Nieto-Hidalgo , Juan M. Espin and more
Potential Business Impact:
Catches fake faces in pictures and videos.
Plain English Summary
Imagine someone creating a fake video of you saying or doing something you never did, which could be used to trick people or steal your identity. This new method is like a super-powered lie detector for videos, able to spot even the most convincing fakes by looking at tiny details that normal methods miss. This means we can trust online videos more and be safer from scams or identity theft.
The rapid advancement of deepfake and face swap technologies has raised significant concerns in digital security, particularly in identity verification and onboarding processes. Conventional detection methods often struggle to generalize against sophisticated facial manipulations. This study proposes an enhanced deep-learning detection framework that combines handcrafted frequency-domain features with conventional RGB inputs. This hybrid approach exploits frequency and spatial domain artifacts introduced during image manipulation, providing richer and more discriminative information to the classifier. Several frequency handcrafted features were evaluated, including the Steganalysis Rich Model, Discrete Cosine Transform, Error Level Analysis, Singular Value Decomposition, and Discrete Fourier Transform
Similar Papers
Towards Generalizable Deepfake Detection with Spatial-Frequency Collaborative Learning and Hierarchical Cross-Modal Fusion
CV and Pattern Recognition
Finds fake videos better, even new kinds.
Towards Sustainable Universal Deepfake Detection with Frequency-Domain Masking
CV and Pattern Recognition
Finds fake pictures even from new AI.
Disruptive Attacks on Face Swapping via Low-Frequency Perceptual Perturbations
CV and Pattern Recognition
Stops fake videos from fooling people.