Rotation invariant CNN using scattering transform for image classification

TL;DR

This paper introduces a rotation-invariant CNN architecture using scattering transforms that accurately predicts input orientation without angle annotations, enhancing robustness in image classification tasks involving rotated data.

Contribution

The paper presents a novel rotation-invariant CNN leveraging scattering transforms and 3D convolutions, capable of predicting orientations continuously without angle labels.

Findings

Achieves rotation invariance in image classification

Predicts continuous orientation angles

Effective with randomly rotated training data

Abstract

Deep convolutional neural networks accuracy is heavily impacted by rotations of the input data. In this paper, we propose a convolutional predictor that is invariant to rotations in the input. This architecture is capable of predicting the angular orientation without angle-annotated data. Furthermore, the predictor maps continuously the random rotation of the input to a circular space of the prediction. For this purpose, we use the roto-translation properties existing in the Scattering Transform Networks with a series of 3D Convolutions. We validate the results by training with upright and randomly rotated samples. This allows further applications of this work on fields like automatic re-orientation of randomly oriented datasets.