Examining the Impact of Blur on Recognition by Convolutional Networks

TL;DR

This paper studies how optical blur affects CNN-based recognition, showing that training with blurred images improves robustness and that models learn blur-invariant features, enhancing real-world recognition performance.

Contribution

It demonstrates that fine-tuning CNNs with blurred images enhances robustness and reveals that models learn to generate blur-invariant representations in hidden layers.

Findings

Fine-tuning with blurred images recovers lost accuracy.

Models generalize across different types and degrees of blur.

Blur-invariant features emerge in hidden layers.

Abstract

State-of-the-art algorithms for many semantic visual tasks are based on the use of convolutional neural networks. These networks are commonly trained, and evaluated, on large annotated datasets of artifact-free high-quality images. In this paper, we investigate the effect of one such artifact that is quite common in natural capture settings: optical blur. We show that standard network models, trained only on high-quality images, suffer a significant degradation in performance when applied to those degraded by blur due to defocus, or subject or camera motion. We investigate the extent to which this degradation is due to the mismatch between training and input image statistics. Specifically, we find that fine-tuning a pre-trained model with blurred images added to the training set allows it to regain much of the lost accuracy. We also show that there is a fair amount of generalization between different degrees and types of blur, which implies that a single network model can be used robustly for recognition when the nature of the blur in the input is unknown. We find that this robustness arises as a result of these models learning to generate blur invariant representations in their hidden layers. Our findings provide useful insights towards developing vision systems that can perform reliably on real world images affected by blur.