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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Accurately and efficiently extracting building footprints from a wide rangeof remote sensed imagery remains a challenge due to their complex structure,variety of scales and diverse appearances. Existing convolutional neuralnetwork (CNN)-based building extraction methods are complained that they cannotdetect the tiny buildings because the spatial information of CNN feature mapsare lost during repeated pooling operations of the CNN, and the large buildingsstill have inaccurate segmentation edges. Moreover, features extracted by a CNNare always partial which restricted by the size of the respective field, andlarge-scale buildings with low texture are always discontinuous and holey whenextracted. This paper proposes a novel multi attending path neural network(MAP-Net) for accurately extracting multiscale building footprints and preciseboundaries. MAP-Net learns spatial localization-preserved multiscale featuresthrough a multi-parallel path in which each stage is gradually generated toextract high-level semantic features with fixed resolution. Then, an attentionmodule adaptively squeezes channel-wise features from each path foroptimization, and a pyramid spatial pooling module captures global dependencyfor refining discontinuous building footprints. Experimental results show thatMAP-Net outperforms state-of-the-art (SOTA) algorithms in boundary localizationaccuracy as well as continuity of large buildings. Specifically, our methodachieved 0.68\%, 1.74\%, 1.46\% precision, and 1.50\%, 1.53\%, 0.82\% IoU scoreimprovement without increasing computational complexity compared with thelatest HRNetv2 on the Urban 3D, Deep Globe and WHU datasets, respectively. TheTensorFlow implementation is available at https://github.com/lehaifeng/MAPNet.

MAP-Net: Multiple Attending Path Neural Network for Building Footprint Extraction From Remote Sensed Imagery