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

2025-06-13

AI Summary

To address the significant performance degradation of synthetic image detection under realistic internet-induced distortions—such as compression and scaling—this paper proposes a robust synthetic image detection framework. The method introduces three key contributions: (1) a genetic algorithm-based automated augmentation policy search that jointly optimizes for discriminability and invariance; (2) a multi-stage data augmentation pipeline coupled with a dual-objective loss function, explicitly enhancing model robustness against quality degradation and geometric transformations; and (3) an end-to-end detector built upon fine-tuned ResNet-50. Evaluated on a benchmark dataset comprising perturbed synthetic images, the proposed approach achieves a 22.53% average precision improvement over strong baselines, outperforming existing state-of-the-art methods. The source code is publicly available.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Search and Optimization: Learning to Search
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Responsible Web: Machine-in-the-loop, human agency and autonomy

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.

Problem

• Detect synthetic images altered by real-world perturbations
• Improve Synthetic Image Detection (SID) with data augmentation
• Enhance model performance across varying image transformations

Innovation

• Genetic algorithm for optimal augmentation selection
• Dual-criteria optimization approach
• Composite data augmentations for robustness