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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Segmenting aerial images is being of great potential in surveillance andscene understanding of urban areas. It provides a mean for automatic reportingof the different events that happen in inhabited areas. This remarkablypromotes public safety and traffic management applications. After the wideadoption of convolutional neural networks methods, the accuracy of semanticsegmentation algorithms could easily surpass 80% if a robust dataset isprovided. Despite this success, the deployment of a pre-trained segmentationmodel to survey a new city that is not included in the training setsignificantly decreases the accuracy. This is due to the domain shift betweenthe source dataset on which the model is trained and the new target domain ofthe new city images. In this paper, we address this issue and consider thechallenge of domain adaptation in semantic segmentation of aerial images. Wedesign an algorithm that reduces the domain shift impact using GenerativeAdversarial Networks (GANs). In the experiments, we test the proposedmethodology on the International Society for Photogrammetry and Remote Sensing(ISPRS) semantic segmentation dataset and found that our method improves theoverall accuracy from 35% to 52% when passing from Potsdam domain (consideredas source domain) to Vaihingen domain (considered as target domain). Inaddition, the method allows recovering efficiently the inverted classes due tosensor variation. In particular, it improves the average segmentation accuracyof the inverted classes due to sensor variation from 14% to 61%.

Unsupervised Domain Adaptation using Generative Adversarial Networks for Semantic Segmentation of Aerial Images