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

This paper addresses domain adaptation for the pixel-wise classification ofremotely sensed data using deep neural networks (DNN) as a strategy to reducethe requirements of DNN with respect to the availability of training data. Wefocus on the setting in which labelled data are only available in a sourcedomain DS, but not in a target domain DT. Our method is based on adversarialtraining of an appearance adaptation network (AAN) that transforms images fromDS such that they look like images from DT. Together with the original labelmaps from DS, the transformed images are used to adapt a DNN to DT. We proposea joint training strategy of the AAN and the classifier, which constrains theAAN to transform the images such that they are correctly classified. In thisway, objects of a certain class are changed such that they resemble objects ofthe same class in DT. To further improve the adaptation performance, we proposea new regularization loss for the discriminator network used in domainadversarial training. We also address the problem of finding the optimal valuesof the trained network parameters, proposing an unsupervised entropy basedparameter selection criterion which compensates for the fact that there is novalidation set in DT that could be monitored. As a minor contribution, wepresent a new weighting strategy for the cross-entropy loss, addressing theproblem of imbalanced class distributions. Our method is evaluated in 42adaptation scenarios using datasets from 7 cities, all consisting ofhigh-resolution digital orthophotos and height data. It achieves a positivetransfer in all cases, and on average it improves the performance in the targetdomain by 4.3% in overall accuracy. In adaptation scenarios between datasetsfrom the ISPRS semantic labelling benchmark our method outperforms those fromrecent publications by 10-20% with respect to the mean intersection over union.

Appearance Based Deep Domain Adaptation for the Classification of Aerial Images