Data-Driven Deepfake Image Detection Method -- The 2024 Global Deepfake Image Detection Challenge

TL;DR

This paper presents a deepfake detection method using Swin Transformer V2-B, enhanced by data augmentation and sample generation, achieving high accuracy in the 2024 Global Deepfake Image Detection Challenge.

Contribution

It introduces a novel deepfake detection approach leveraging Swin Transformer V2-B with data augmentation techniques to improve generalization.

Findings

Achieved excellence award in deepfake detection challenge.

Enhanced model robustness through data augmentation.

Demonstrated effectiveness of Transformer-based architecture.

Abstract

With the rapid development of technology in the field of AI, deepfake technology has emerged as a double-edged sword. It has not only created a large amount of AI-generated content but also posed unprecedented challenges to digital security. The task of the competition is to determine whether a face image is a Deepfake image and output its probability score of being a Deepfake image. In the image track competition, our approach is based on the Swin Transformer V2-B classification network. And online data augmentation and offline sample generation methods are employed to enrich the diversity of training samples and increase the generalization ability of the model. Finally, we got the award of excellence in Deepfake image detection.