Cut-Thumbnail: A Novel Data Augmentation for Convolutional Neural Network

TL;DR

This paper introduces Cut-Thumbnail, a new data augmentation method that enhances shape bias in CNNs by replacing image regions with reduced thumbnails, improving performance across multiple vision tasks.

Contribution

The paper presents a novel augmentation strategy combining image reduction and region replacement, integrated with Mixup, to improve CNN shape bias and accuracy.

Findings

Outperforms state-of-the-art augmentation methods on various tasks.

Achieves 79.21% accuracy on ImageNet with ResNet-50, surpassing baseline by 2.8%.

Effective in classification, fine-grained classification, and object detection.

Abstract

In this paper, we propose a novel data augmentation strategy named Cut-Thumbnail, that aims to improve the shape bias of the network. We reduce an image to a certain size and replace the random region of the original image with the reduced image. The generated image not only retains most of the original image information but also has global information in the reduced image. We call the reduced image as thumbnail. Furthermore, we find that the idea of thumbnail can be perfectly integrated with Mixed Sample Data Augmentation, so we put one image's thumbnail on another image while the ground truth labels are also mixed, making great achievements on various computer vision tasks. Extensive experiments show that Cut-Thumbnail works better than state-of-the-art augmentation strategies across classification, fine-grained image classification, and object detection. On ImageNet classification, ResNet-50 architecture with our method achieves 79.21\% accuracy, which is more than 2.8\% improvement on the baseline.