Deepfake Detection: A Comprehensive Survey from the Reliability Perspective

View on arXiv ← Back to list

Authors: Tianyi Wang, Xin Liao, Kam Pui Chow, Xiaodong Lin, Yinglong Wang

Published: 2022-11-20 06:31:23+00:00

AI Summary

This survey reviews existing deepfake detection studies from a reliability perspective, identifying key challenges in transferability, interpretability, and robustness. The authors introduce a model reliability study metric using statistical random sampling to evaluate the reliability of detection models and provide case studies on real-life deepfakes.

Abstract

The mushroomed Deepfake synthetic materials circulated on the internet have raised a profound social impact on politicians, celebrities, and individuals worldwide. In this survey, we provide a thorough review of the existing Deepfake detection studies from the reliability perspective. We identify three reliability-oriented research challenges in the current Deepfake detection domain: transferability, interpretability, and robustness. Moreover, while solutions have been frequently addressed regarding the three challenges, the general reliability of a detection model has been barely considered, leading to the lack of reliable evidence in real-life usages and even for prosecutions on Deepfake-related cases in court. We, therefore, introduce a model reliability study metric using statistical random sampling knowledge and the publicly available benchmark datasets to review the reliability of the existing detection models on arbitrary Deepfake candidate suspects. Case studies are further executed to justify the real-life Deepfake cases including different groups of victims with the help of the reliably qualified detection models as reviewed in this survey. Reviews and experiments on the existing approaches provide informative discussions and future research directions for Deepfake detection.

Key findings

The study reveals trade-offs between transferability, interpretability, and robustness in existing deepfake detection models. No single model consistently outperforms others across all datasets. A model reliability study scheme helps quantify the trustworthiness of detection models for real-world applications and legal proceedings.

Approach

The paper proposes a model reliability study metric using statistical random sampling to assess the reliability of deepfake detection models. This involves constructing a population from available datasets, performing random sampling, and computing confidence intervals for accuracy and AUC score metrics.

Datasets

FaceForensics++, Deepfake Detection Challenge (DFDC), Celeb-DF, DeeperForensics-1.0 (DF1.0), FaceShifter

Model(s)

Xception, MAT, RECCE, Stage5, FSTMatching, MetricLearning, LRNet

Author countries

China, Canada

← Previous