MoireDB: Formula-generated Interference-fringe Image Dataset

TL;DR

The paper addresses the challenge of improving image-classification robustness to real-world degradations by proposing MoireDB, a formula-generated dataset of interference-fringe (Moiré) images for PixMix-style augmentation. MoireDB generates images from mathematical formulas to avoid copyright and cost issues associated with fractal and FVis sources, aiming to match or surpass their effectiveness. Experimental results on CIFAR-C show that MoireDB-based augmentation yields robust performance improvements across multiple corruption and adversarial settings, often outperforming Fractal arts and FVis augmentations. The work demonstrates that illusory, formula-generated images can be a scalable, privacy-safe augmentation strategy with practical impact for deploying robust vision models.

Abstract

Image recognition models have struggled to treat recognition robustness to real-world degradations. In this context, data augmentation methods like PixMix improve robustness but rely on generative arts and feature visualizations (FVis), which have copyright, drawing cost, and scalability issues. We propose MoireDB, a formula-generated interference-fringe image dataset for image augmentation enhancing robustness. MoireDB eliminates copyright concerns, reduces dataset assembly costs, and enhances robustness by leveraging illusory patterns. Experiments show that MoireDB augmented images outperforms traditional Fractal arts and FVis-based augmentations, making it a scalable and effective solution for improving model robustness against real-world degradations.

Figures (5)

• Figure 1: Data augmentation methods based on formula-generated images ImageNet-CPIXMIX.
• Figure 2: CIFAR-C ImageNet-C.
• Figure 3: Examples of data augmentation images PIXMIXFractalDBVisualAtom.
• Figure 4: Our algorithm for generating Moiré images.
• Figure 5: Example illustrating PIXMIX-style data augmentation with Moiré images ImageNet-CPIXMIX.