Real-Time Deepfake Detection in the Real-World

TL;DR

This paper presents LaDeDa, a patch-based deepfake detection algorithm with near-perfect accuracy on benchmarks, and introduces WildRF, a new real-world dataset revealing the ongoing challenge of generalizing deepfake detection to social media content.

Contribution

The paper introduces LaDeDa, a highly accurate patch-based deepfake detector, and WildRF, a new dataset for real-world deepfake detection, highlighting the gap between benchmark performance and real-world applicability.

Findings

LaDeDa achieves around 99% mAP on current benchmarks.

Tiny-LaDeDa is highly efficient with 375x fewer FLOPs.

Performance drops significantly on real-world social media deepfakes.

Abstract

Recent improvements in generative AI made synthesizing fake images easy; as they can be used to cause harm, it is crucial to develop accurate techniques to identify them. This paper introduces "Locally Aware Deepfake Detection Algorithm" (LaDeDa), that accepts a single 9x9 image patch and outputs its deepfake score. The image deepfake score is the pooled score of its patches. With merely patch-level information, LaDeDa significantly improves over the state-of-the-art, achieving around 99% mAP on current benchmarks. Owing to the patch-level structure of LaDeDa, we hypothesize that the generation artifacts can be detected by a simple model. We therefore distill LaDeDa into Tiny-LaDeDa, a highly efficient model consisting of only 4 convolutional layers. Remarkably, Tiny-LaDeDa has 375x fewer FLOPs and is 10,000x more parameter-efficient than LaDeDa, allowing it to run efficiently on edge devices with a minor decrease in accuracy. These almost-perfect scores raise the question: is the task of deepfake detection close to being solved? Perhaps surprisingly, our investigation reveals that current training protocols prevent methods from generalizing to real-world deepfakes extracted from social media. To address this issue, we introduce WildRF, a new deepfake detection dataset curated from several popular social networks. Our method achieves the top performance of 93.7% mAP on WildRF, however the large gap from perfect accuracy shows that reliable real-world deepfake detection is still unsolved.