FFR_FD: Effective and Fast Detection of DeepFakes Based on Feature Point Defects

View on arXiv ← Back to list

Authors: Gaojian Wang, Qian Jiang, Xin Jin, Xiaohui Cui

Published: 2021-07-05 13:35:39+00:00

AI Summary

This paper proposes FFR_FD, a novel method for fast and effective deepfake detection. FFR_FD leverages the observation that deepfakes have fewer feature points than real faces, especially in specific facial regions, to create a compact feature vector for classification using a Random Forest.

Abstract

The internet is filled with fake face images and videos synthesized by deep generative models. These realistic DeepFakes pose a challenge to determine the authenticity of multimedia content. As countermeasures, artifact-based detection methods suffer from insufficiently fine-grained features that lead to limited detection performance. DNN-based detection methods are not efficient enough, given that a DeepFake can be created easily by mobile apps and DNN-based models require high computational resources. For the first time, we show that DeepFake faces have fewer feature points than real ones, especially in certain facial regions. Inspired by feature point detector-descriptors to extract discriminative features at the pixel level, we propose the Fused Facial Region_Feature Descriptor (FFR_FD) for effective and fast DeepFake detection. FFR_FD is only a vector extracted from the face, and it can be constructed from any feature point detector-descriptors. We train a random forest classifier with FFR_FD and conduct extensive experiments on six large-scale DeepFake datasets, whose results demonstrate that our method is superior to most state of the art DNN-based models.

Key findings

The proposed FFR_FD method outperforms most state-of-the-art DNN-based models in deepfake detection across several benchmark datasets. The method demonstrates effectiveness and efficiency, using a lightweight model trained on a CPU, and shows good generalization capability across different deepfake datasets.

Approach

The approach extracts feature points from eight facial regions using various detectors (SIFT, SURF, FAST, ORB, A-KAZE). The descriptors of these points are fused within each region to create a compact vector (FFR_FD), which is then used to train a Random Forest classifier for deepfake detection.

Datasets

DeepfakeTIMIT (HQ and LQ), UADFV, FF++ DeepFakes (RAW and LQ), DFD, DFDC, CelebDF V2

Model(s)

Random Forest

Author countries

China

← Previous