English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Computer vision using deep neural networks (DNNs) has brought about seminalchanges in people's lives. Applications range from automotive, face recognitionin the security industry, to industrial process monitoring. In some cases, DNNsinfer even in safety-critical situations. Therefore, for practicalapplications, DNNs have to behave in a robust way to disturbances such asnoise, pixelation, or blur. Blur directly impacts the performance of DNNs,which are often approximated as a disk-shaped kernel to model defocus. However,optics suggests that there are different kernel shapes depending on wavelengthand location caused by optical aberrations. In practice, as the optical qualityof a lens decreases, such aberrations increase. This paper proposesOpticsBench, a benchmark for investigating robustness to realistic, practicallyrelevant optical blur effects. Each corruption represents an optical aberration(coma, astigmatism, spherical, trefoil) derived from Zernike Polynomials.Experiments on ImageNet show that for a variety of different pre-trained DNNs,the performance varies strongly compared to disk-shaped kernels, indicating thenecessity of considering realistic image degradations. In addition, we show onImageNet-100 with OpticsAugment that robustness can be increased by usingoptical kernels as data augmentation. Compared to a conventionally trainedResNeXt50, training with OpticsAugment achieves an average performance gain of21.7% points on OpticsBench and 6.8% points on 2D common corruptions.

Classification robustness to common optical aberrations