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/

Zendy Plus

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

Zendy Tools

Zendy Open

Convolutional neural networks (ConvNets) have been successfully applied tosatellite image scene classification. Human-labeled training datasets areessential for ConvNets to perform accurate classification. Errors inhuman-annotated training datasets are unavoidable due to the complexity ofsatellite images. However, the distribution of real-world human-annotated labelnoises on remote sensing images and their impact on ConvNets have not beeninvestigated. To fill this research gap, this study, for the first time,collected real-world labels from 32 participants and explored how theirannotated label noise affect three representative ConvNets (VGG16, GoogleNet,and ResNet-50) for remote sensing image scene classification. We found that:(1) human-annotated label noise exhibits significant class and instancedependence; (2) an additional 1% of human-annotated label noise in trainingdata leads to 0.5% reduction in the overall accuracy of ConvNetsclassification; (3) the error pattern of ConvNet predictions was stronglycorrelated with that of participant's labels. To uncover the mechanismunderlying the impact of human labeling errors on ConvNets, we further comparedit with three types of simulated label noise: uniform noise, class-dependentnoise and instance-dependent noise. Our results show that the impact ofhuman-annotated label noise on ConvNets significantly differs from all threetypes of simulated label noise, while both class dependence and instancedependence contribute to the impact of human-annotated label noise on ConvNets.These observations necessitate a reevaluation of the handling of noisy labels,and we anticipate that our real-world label noise dataset would facilitate thefuture development and assessment of label-noise learning algorithms.

Human-annotated label noise and their impact on ConvNets for remote sensing image scene classification