Detect Any Deepfakes

By Y. Lai et al
Published on June 29, 2023
Read the original document by opening this link in a new tab.

Table of Contents

1 Introduction
2 Related work
2.1 Face Forgery Detection
2.2 Foundation Model
2.3 Parameter-Efficient Fine-Tuning
3 Methodology
3.1 Multiscale Adapter for SAM
3.2 Reconstruction Guided Attention
3.3 Loss Function
4 Experiments
4.1 Datasets and Performance Metrics

Summary

The paper 'Detect Any Deepfakes' introduces a framework called DADF for face forgery detection and localization. The framework utilizes the Segment Anything Model (SAM) with a Multiscale Adapter and Reconstruction Guided Attention module. The authors address the challenges of precise forgery detection and localization by proposing novel methods. Extensive experiments on benchmark datasets demonstrate the superiority of the approach for both forgery detection and localization.
×
This is where the content will go.