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

Land cover classification is a multi-class segmentation task to classify eachpixel into a certain natural or man-made category of the earth surface, such aswater, soil, natural vegetation, crops, and human infrastructure. Limited byhardware computational resources and memory capacity, most existing studiespreprocessed original remote sensing images by down sampling or cropping theminto small patches less than 512*512 pixels before sending them to a deepneural network. However, down sampling images incurs spatial detail loss,renders small segments hard to discriminate, and reverses the spatialresolution progress obtained by decades of years of efforts. Cropping imagesinto small patches causes a loss of long-range context information, andrestoring the predicted results to their original size brings extra latency. Inresponse to the above weaknesses, we present an efficient lightweight semanticsegmentation network termed MKANet. Aimed at the characteristics of top viewhigh-resolution remote sensing imagery, MKANet utilizes sharing kernels tosimultaneously and equally handle ground segments of inconsistent scales, andalso employs parallel and shallow architecture to boost inference speed andfriendly support image patches more than 10X larger. To enhance boundary andsmall segments discrimination, we also propose a method that captures categoryimpurity areas, exploits boundary information and exerts an extra penalty onboundaries and small segment misjudgment. Both visual interpretations andquantitative metrics of extensive experiments demonstrate that MKANet acquiresstate-of-the-art accuracy on two land-cover classification datasets and infers2X faster than other competitive lightweight networks. All these meritshighlight the potential of MKANet in practical applications.

MKANet: A Lightweight Network with Sobel Boundary Loss for Efficient Land-cover Classification of Satellite Remote Sensing Imagery