# LocalNorm: Robust Image Classification through Dynamically Regularized Normalization

## TL;DR

LocalNorm is a novel normalization technique for CNNs that enhances robustness to image noise by dynamically adapting to local image properties, significantly improving degraded image classification without substantial computational overhead.

## Contribution

Introduces LocalNorm, a dynamic normalization method that improves CNN robustness to image degradation with minimal additional training or inference costs.

## Key findings

- Increases noise robustness up to three times
- Maintains or slightly improves accuracy on standard benchmarks
- Adds negligible computational overhead

## Abstract

While modern convolutional neural networks achieve outstanding accuracy on many image classification tasks, they are, compared to humans, much more sensitive to image degradation. Here, we describe a variant of Batch Normalization, LocalNorm, that regularizes the normalization layer in the spirit of Dropout while dynamically adapting to the local image intensity and contrast at test-time. We show that the resulting deep neural networks are much more resistant to noise-induced image degradation, improving accuracy by up to three times, while achieving the same or slightly better accuracy on non-degraded classical benchmarks. In computational terms, LocalNorm adds negligible training cost and little or no cost at inference time, and can be applied to already-trained networks in a straightforward manner.