Geometric Data Augmentation Based on Feature Map Ensemble

TL;DR

This paper introduces a novel CNN architecture that enhances robustness to geometric transformations by enclosing existing backbones with geometric transformations and feature map ensemble, improving performance on standard datasets.

Contribution

It proposes a new CNN design that improves geometric transformation robustness without altering existing backbones, compatible with current data augmentation methods.

Findings

Improved accuracy on CIFAR, CUB-200, and Mnist-rot-12k datasets.

Enhanced robustness against large rotations.

Compatible with state-of-the-art data augmentation techniques.

Abstract

Deep convolutional networks have become the mainstream in computer vision applications. Although CNNs have been successful in many computer vision tasks, it is not free from drawbacks. The performance of CNN is dramatically degraded by geometric transformation, such as large rotations. In this paper, we propose a novel CNN architecture that can improve the robustness against geometric transformations without modifying the existing backbones of their CNNs. The key is to enclose the existing backbone with a geometric transformation (and the corresponding reverse transformation) and a feature map ensemble. The proposed method can inherit the strengths of existing CNNs that have been presented so far. Furthermore, the proposed method can be employed in combination with state-of-the-art data augmentation algorithms to improve their performance. We demonstrate the effectiveness of the proposed method using standard datasets such as CIFAR, CUB-200, and Mnist-rot-12k.