Lightweight Deepfake Detection Based on Multi-Feature Fusion

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(4), P. 1954 - 1954

Published: Feb. 13, 2025

Deepfake technology utilizes deep learning (DL)-based face manipulation techniques to seamlessly replace faces in videos, creating highly realistic but artificially generated content. Although this has beneficial applications media and entertainment, misuse of its capabilities may lead serious risks, including identity theft, cyberbullying, false information. The integration DL with visual cognition resulted important technological improvements, particularly addressing privacy risks caused by “deepfake” images on digital platforms. In study, we propose an efficient lightweight method for detecting deepfake making it suitable devices limited computational resources. order reduce the burden usually associated models, our integrates machine classifiers combination keyframing approaches texture analysis. Moreover, features extracted a histogram oriented gradients (HOG), local binary pattern (LBP), KAZE bands were integrated evaluate using random forest, extreme gradient boosting, extra trees, support vector classifier algorithms. Our findings show feature-level fusion HOG, LBP, improves accuracy 92% 96% FaceForensics++ Celeb-DF(v2), respectively.

Language: Английский

A survey on multimedia-enabled deepfake detection: state-of-the-art tools and techniques, emerging trends, current challenges & limitations, and future directions

Deleted Journal, Journal Year: 2025, Volume and Issue: 28(1)

Published: April 25, 2025

Language: Английский