Local Magnification for Data and Feature Augmentation

TL;DR

This paper introduces Local Magnification (LOMA), a simple, model-free data augmentation technique that enhances training data by locally magnifying image regions, improving classification and detection accuracy.

Contribution

The paper proposes LOMA, a novel local magnification augmentation method, and extends it to feature space, outperforming existing augmentation techniques in image classification and object detection.

Findings

LOMA significantly improves classification accuracy.

Combining LOMA with feature augmentation further boosts performance.

LOMA outperforms advanced intensity transformation methods.

Abstract

In recent years, many data augmentation techniques have been proposed to increase the diversity of input data and reduce the risk of overfitting on deep neural networks. In this work, we propose an easy-to-implement and model-free data augmentation method called Local Magnification (LOMA). Different from other geometric data augmentation methods that perform global transformations on images, LOMA generates additional training data by randomly magnifying a local area of the image. This local magnification results in geometric changes that significantly broaden the range of augmentations while maintaining the recognizability of objects. Moreover, we extend the idea of LOMA and random cropping to the feature space to augment the feature map, which further boosts the classification accuracy considerably. Experiments show that our proposed LOMA, though straightforward, can be combined with standard data augmentation to significantly improve the performance on image classification and object detection. And further combination with our feature augmentation techniques, termed LOMA_IF&FO, can continue to strengthen the model and outperform advanced intensity transformation methods for data augmentation.