Composite Data Augmentations for Synthetic Image Detection Against Real-World Perturbations

TL;DR

This paper proposes a novel data augmentation strategy using genetic algorithms and dual-criteria optimization to enhance synthetic image detection robustness against real-world perturbations, significantly improving model performance.

Contribution

It introduces a composite data augmentation approach optimized via genetic algorithms and dual-criteria methods to improve synthetic image detection under diverse real-world conditions.

Findings

Mean average precision increased by +22.53% with the proposed method.

Enhanced detection robustness against compression and other perturbations.

Provides insights for developing more resilient synthetic image detection models.

Abstract

The advent of accessible Generative AI tools enables anyone to create and spread synthetic images on social media, often with the intention to mislead, thus posing a significant threat to online information integrity. Most existing Synthetic Image Detection (SID) solutions struggle on generated images sourced from the Internet, as these are often altered by compression and other operations. To address this, our research enhances SID by exploring data augmentation combinations, leveraging a genetic algorithm for optimal augmentation selection, and introducing a dual-criteria optimization approach. These methods significantly improve model performance under real-world perturbations. Our findings provide valuable insights for developing detection models capable of identifying synthetic images across varying qualities and transformations, with the best-performing model achieving a mean average precision increase of +22.53% compared to models without augmentations. The implementation is available at github.com/efthimia145/sid-composite-data-augmentation.