This global dataset captures a picture of Earth in the smallest detail ever


تلتقط مجموعة البيانات العالمية صورة الأرض بأدق التفاصيل على الإطلاق arXiv (2022). DOI: 10.48550 / arxiv.2207.06418″ width=”800″ height=”530″/>

Summarize the syntax and classes of the WorldStrat dataset. attributed to him: arXiv (2022). DOI: 10.48550/arxiv.2207.06418

An open source global dataset of high-resolution images of the Earth – the most comprehensive and detailed of its kind – was developed by experts led by UCL with data from the European Space Agency (ESA).


The free dataset, WorldStrat, will be presented at the NeurIPS 2022 conference in New Orleans. It includes approximately 10,000 square kilometers of free space Satellites Pictures, showing each type of location, urban area And the Land use From agriculture, grasslands and forests to cities of all sizes and polar ice caps.

The dataset includes locations in the Global South and those you need Humanitarian aidwhich is often underrepresented in satellite images Because this is usually collected for commercial gain, and so it is disproportionately characterized by wealthier regions.

Scientists say the combination enables global analysis of terrain to address global challenges such as natural and man-made disaster response, natural resource management and urban planning.

It started working on WorldStrat in 2021, and since its launch in June 2022 it has been downloaded more than 3,000 times.

Project leader Dr Julian Kornepez (UCL Computer Science) said, “Combining high-resolution commercial imagery and… machine learning It has huge potential to enable planet-wide analytics, which can help tackle all kinds of global challenges — the problem is that business data is often locked behind a paywall. “

“ESA’s TPM program made our project possible by providing it free access to data that is usually very expensive.”

The team used data from the Airbus SPOT 6 and SPOT 7 satellites, which were commissioned by the European Space Agency and launched in 2012 and 2014, respectively. Satellites can provide images with a resolution of up to 1.5 meters per pixel, which means that each pixel represents an area of ​​1.5 meters by 1.5 meters on Earth.

The scientists used about 4,000 highly detailed images from the SPOT satellite. Even those with high (spatial) resolution, they have low temporal resolution, which in this context means that not every satellite revisits and retrieves every site regularly. This is because the images captured by the satellites were originally intended to be used for specific commercial applications rather than for long-term analyses.

To combat this, the team also used freely available low-resolution images from the Copernicus Sentinel-2 satellite. This is higher Temporal resolution, which means that they were caught at more regular time points every five days. They matched each SPOT image against 16 Copernicus Sentinel-2 images, using about 64,000 in all.

The researchers developed the dataset to support the development of machine learning applications to extend and enhance it, for example to further improve image resolution. To allow more applications to be developed, scientists have developed an AI toolkit in addition to the full source code, allowing developers to reproduce, extend, and transform the work.

“Thousands of data users from around the world have already downloaded WorldStrat – and we look forward to seeing ways they can extend and improve it using machine learning techniques,” Dr. Kornepez continued.

A printed copy of the research is available at arXiv.

more information:
Julian Kornepez et al., Unlocking High-Resolution Satellite Imagery: The WorldStrat Data Set – With Super Resolution Applied, arXiv (2022). DOI: 10.48550/arxiv.2207.06418

GitHub dataset: worldstrat.github.io/

Journal information:
arXiv

the quote: Global Dataset Captures Earth in Finest Detail Ever (2022, November 18) Retrieved November 18, 2022 from https://phys.org/news/2022-11-worldwide-dataset-captures-earth-finest.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.





Source link

Related Posts

Precaliga