Motion Equivariance OF Event-based Camera Data with the Temporal Normalization Transform

TL;DR

This paper investigates how to achieve motion equivariance in event-based camera data using the Temporal Normalization Transform to improve CNN robustness to motion variations in object recognition tasks.

Contribution

It introduces a novel approach leveraging the Temporal Normalization Transform to enhance CNNs' motion equivariance for event-based data.

Findings

Improved robustness of CNNs to motion variations in event data

Demonstrated effectiveness of the Temporal Normalization Transform

Enhanced object recognition accuracy under diverse motions

Abstract

In this work, we focus on using convolution neural networks (CNN) to perform object recognition on the event data. In object recognition, it is important for a neural network to be robust to the variations of the data during testing. For traditional cameras, translations are well handled because CNNs are naturally equivariant to translations. However, because event cameras record the change of light intensity of an image, the geometric shape of event volumes will not only depend on the objects but also on their relative motions with respect to the camera. The deformation of the events caused by motions causes the CNN to be less robust to unseen motions during inference. To address this problem, we would like to explore the equivariance property of CNNs, a well-studied area that demonstrates to produce predictable deformation of features under certain transformations of the input image.