A precortical module for robust CNNs to light variations

TL;DR

This paper introduces a biologically inspired precortical module for CNNs, modeled after mammalian visual pathways, which enhances robustness to light intensity and contrast variations in image classification tasks.

Contribution

The paper proposes a novel precortical module inspired by mammalian visual circuits, integrated into CNNs to improve robustness against lighting and contrast changes.

Findings

Significantly increased robustness to light and contrast variations.

Effective on MNIST, FashionMNIST, and SVHN datasets.

Simple addition improves CNN performance under varying lighting conditions.

Abstract

We present a simple mathematical model for the mammalian low visual pathway, taking into account its key elements: retina, lateral geniculate nucleus (LGN), primary visual cortex (V1). The analogies between the cortical level of the visual system and the structure of popular CNNs, used in image classification tasks, suggests the introduction of an additional preliminary convolutional module inspired to precortical neuronal circuits to improve robustness with respect to global light intensity and contrast variations in the input images. We validate our hypothesis on the popular databases MNIST, FashionMNIST and SVHN, obtaining significantly more robust CNNs with respect to these variations, once such extra module is added.