Data Augmentation for Image Classification using Generative AI

TL;DR

This paper introduces AGA, a novel data augmentation framework leveraging generative AI to enhance image classification datasets by preserving foregrounds and diversifying backgrounds, leading to significant accuracy improvements.

Contribution

The paper presents a new augmentation method combining LLMs, diffusion, and segmentation models to improve dataset diversity while maintaining subject authenticity.

Findings

15.6% accuracy increase on in-distribution data

23.5% accuracy increase on out-of-distribution data

64.3% SIC score improvement over baselines

Abstract

Scaling laws dictate that the performance of AI models is proportional to the amount of available data. Data augmentation is a promising solution to expanding the dataset size. Traditional approaches focused on augmentation using rotation, translation, and resizing. Recent approaches use generative AI models to improve dataset diversity. However, the generative methods struggle with issues such as subject corruption and the introduction of irrelevant artifacts. In this paper, we propose the Automated Generative Data Augmentation (AGA). The framework combines the utility of large language models (LLMs), diffusion models, and segmentation models to augment data. AGA preserves foreground authenticity while ensuring background diversity. Specific contributions include: i) segment and superclass based object extraction, ii) prompt diversity with combinatorial complexity using prompt decomposition, and iii) affine subject manipulation. We evaluate AGA against state-of-the-art (SOTA) techniques on three representative datasets, ImageNet, CUB, and iWildCam. The experimental evaluation demonstrates an accuracy improvement of 15.6% and 23.5% for in and out-of-distribution data compared to baseline models, respectively. There is also a 64.3% improvement in SIC score compared to the baselines.