Lightweight Deepfake Detection Based on Multi-Feature Fusion

Authors: Siddiqui Muhammad Yasir, Hyun Kim

Published: 2025-02-17 12:55:41+00:00

Journal Ref: Yasir, S.M.; Kim, H. Lightweight Deepfake Detection Based on Multi-Feature Fusion. Appl. Sci. 2025, 15, 1954

AI Summary

This study proposes an efficient and lightweight method for detecting deepfake images and videos, making it suitable for devices with limited computational resources. The approach integrates machine learning classifiers with keyframing and texture analysis, specifically fusing features extracted using Histogram of Oriented Gradients (HOG), Local Binary Pattern (LBP), and KAZE bands. The method demonstrates improved accuracy on standard deepfake datasets while reducing computational burden.

Abstract

Deepfake technology utilizes deep learning based face manipulation techniques to seamlessly replace faces in videos creating highly realistic but artificially generated content. Although this technology has beneficial applications in media and entertainment misuse of its capabilities may lead to serious risks including identity theft cyberbullying and false information. The integration of DL with visual cognition has resulted in important technological improvements particularly in addressing privacy risks caused by artificially generated deepfake images on digital media platforms. In this study we propose an efficient and lightweight method for detecting deepfake images and videos making it suitable for devices with limited computational resources. In order to reduce the computational burden usually associated with DL models our method integrates machine learning classifiers in combination with keyframing approaches and texture analysis. Moreover the features extracted with a histogram of oriented gradients (HOG) local binary pattern (LBP) and KAZE bands were integrated to evaluate using random forest extreme gradient boosting extra trees and support vector classifier algorithms. Our findings show a feature-level fusion of HOG LBP and KAZE features improves accuracy to 92% and 96% on FaceForensics++ and Celeb-DFv2 respectively.

Key findings

The feature-level fusion of HOG, LBP, and KAZE features significantly improved deepfake detection accuracy. The HOG + KAZE fusion achieved 92.12% accuracy on FaceForensics++ and 78% on Celeb-DF using a Support Vector Classifier. This lightweight method offers a good balance between computational efficiency and classification performance, making it suitable for resource-constrained environments.

Approach

The authors extract keyframes from videos, apply texture analysis, and extract features using Histogram of Oriented Gradients (HOG), Local Binary Pattern (LBP), and KAZE descriptors. These features are then fused at the feature level and classified using various machine learning algorithms. The methodology aims for computational efficiency by leveraging traditional ML techniques over complex deep learning models.

Datasets

FaceForensics++, Celeb-DFv2 (referred to as Celeb-DF in the paper body)

Model(s)

Random Forest, Extreme Gradient Boosting (XGBoost), Extra Trees Classifier, Support Vector Classifier (SVC/SVM), integrated with HOG, LBP, and KAZE feature extraction.

Author countries

Republic of Korea

← Previous