LAA-Net: Localized Artifact Attention Network for Quality-Agnostic and Generalizable Deepfake Detection

TL;DR

LAA-Net is a new deepfake detection model that uses explicit attention mechanisms and feature pyramid networks to focus on artifact-prone regions, improving generalization to unseen manipulations.

Contribution

It introduces an explicit attention mechanism within a multi-task framework and an Enhanced Feature Pyramid Network for better artifact localization and feature spreading.

Findings

Outperforms existing methods in AUC and AP metrics

Focuses on small artifact-prone regions for better detection

Demonstrates improved generalization to unseen deepfake manipulations

Abstract

This paper introduces a novel approach for high-quality deepfake detection called Localized Artifact Attention Network (LAA-Net). Existing methods for high-quality deepfake detection are mainly based on a supervised binary classifier coupled with an implicit attention mechanism. As a result, they do not generalize well to unseen manipulations. To handle this issue, two main contributions are made. First, an explicit attention mechanism within a multi-task learning framework is proposed. By combining heatmap-based and self-consistency attention strategies, LAA-Net is forced to focus on a few small artifact-prone vulnerable regions. Second, an Enhanced Feature Pyramid Network (E-FPN) is proposed as a simple and effective mechanism for spreading discriminative low-level features into the final feature output, with the advantage of limiting redundancy. Experiments performed on several benchmarks show the superiority of our approach in terms of Area Under the Curve (AUC) and Average Precision (AP). The code is available at https://github.com/10Ring/LAA-Net.