When data scientist and Linear Logic CEO Miguel Pieters started to apply machine learning algorithms to satellite imaging datasets from Ukraine to assess damage there, he hired a journalist from the country's seaport city of Odessa to navigate the landscape and share insights from the work with local people.
'Naturally, the project goes above and beyond the professional domain,' Sushylova told EE Times in an exclusive interview. 'As a citizen of Ukraine, I assume there is no need to evaluate on how it resonates with me personally-there is no way for it not to. However, despite the whole array of emotions I have been inevitably subject to over the past two years, I aim to place impartiality above everything else. My voice remains my voice, and so do facts.'
Linear Logic, of Amsterdam, is among several satellite and AI companies that have joined in support of Ukraine and coverage of the conflict with open-source imagery and analysis.
By Shingo Kojima, Sr Principal Engineer of Embedded Processing, Renesas Electronics 03.26.2024
Linear Logic's structural damage assessment data on Ukrainian frontline cities is a resource for the Ministry of Infrastructure of Ukraine, and is also used by academic institutions, geographic information system (GIS) practitioners and hobbyists, said Pieters, who is also the founder of Linear Logic, a data-centric platform to promote AI explainability. In addition to its use in damage assessment, the information serves multiple purposes, including reconstructing military advances, assessing infrastructure resilience, identifying vulnerable areas for future planning and evaluating humanitarian aid distribution.
'The data is particularly interesting for areas no longer under Ukrainian control, or directly on the front line, such as Avdiivka, Bakhmut, Mariupol and Vuhledar,' Pieters told EE Times.
Miguel Pieters
For those front-line or occupied cities, satellite imagery is the only way to get complete information of what is happening on the ground. This information is helpful to assess what is needed by any civilians who may still be there, and to make plans to rebuild, Pieters said.
'Moreover, we're proud to share that our data has been instrumental in developing predictive models aimed at forecasting and identifying the likelihood of building damage in urban fighting, further enhancing our ability to anticipate and respond to evolving situations,' he added.
The tech behind the help
Linear Logic has access to satellite imagery with 70cm resolution that covers 85% of Ukraine's total land mass. Its datasets originate from multispectral cameras mounted on satellites from geospatial analytics company Satellogic, of Montevideo, Uruguay, and spatial intelligence platform company Astraea, of Charlottesville, Va.
The results of the analysis in Linear Logic's VartaNet project ('varta' means 'guard' or 'watch' in Ukrainian while 'net' denotes its high-tech quality) are offered as an open-source dataset.
Before the start of Russia's invasion in 2022, as the world debated whether a conflict was imminent, a satellite image from Capella Space, of San Francisco, settled the argument.
Pripyat River crossing
This image, captured through heavy fog and clouds, was taken by Capella Space's synthetic aperture radar technology at the start of the Ukraine conflict, and shows possible evidence of Russian troops crossing the Pripyat River to enter Ukraine. (Source: Capella Space)
The image showed a line of Russian tanks on their side of the border with Ukraine. In the years since, Kyiv's defenses against the Russian Armed Forces have become the first-ever war aided by images taken 375 miles above the earth.
'This is arguably the first large-scale conflict that has been monitored from space with the type of capability and assets that are in space today,' Mack Koepke, VP of global sales for Capella, told EE Times.
Formerly classified tech
Capella's images are from synthetic aperture radar (SAR), a technology developed by commercial interests solely for the U.S. government. Its declassification in 2008 made it available for commercial development.
Analysis by Linear Logic entails a three-tiered damage assessment conducted on a per-object basis. Each building is categorized based on its structural state: undamaged, damaged, or destroyed. Undamaged buildings exhibit no visible damage, while damaged buildings may display partially collapsed roofs or holes in the roof. Destroyed buildings are completely burnt out or collapsed. (Source: Linear Logic)
This is how online publication Defense Advancement explains SAR: 'A remote sensing technology that uses the motion of a radio wave-emitting antenna over a target to create 2D or 3D imagery with greater spatial resolution than conventional radar. The aperture of a radar antenna is usually related to its physical size; the 'synthetic' in SAR refers to the distance that the system moves during the time that a radio-wave pulse is emitted and reflected.'
'In the whole process of doing [SAR] we get wonderful imagery,' Koepke said.
SAR's superpower is that it can produce images even at night or under weather conditions like cloud cover, when optical imagers like Satellogic's do not work.
'Because of that, we were uniquely positioned to help Ukraine when Russia invaded at that time of the year (February) because it's very cloudy during wintertime,' said Koepke.
He went on to describe what was happening behind the scenes at Capella both before and after Russia crossed the border.
'There's a long lead-up to the actual invasion where we had been keeping tabs on that particular area,' he said. 'And that data was being shared with people who needed to see it. Of course, Ukraine themselves, but also allied nations that were very interested in understanding what this meant for regional stability.'
Once the invasion took place, different parties who had a stake in Ukraine started directing Capella to hot spots in the conflict where it should direct its radio waves. Once captured, the space tech company wanted to be sure its images were seen by a world clamoring for news.
'In the early days of that conflict, there was all this data that was being created, but it wasn't necessarily getting into the hands of the people who were very interested in seeing it or needed to see it,' Koepke said. 'So, not only is that people who are on the ground, but it's also media, people wanting to make sure that there's broad exposure of what's happening.'
The media, Ukraine and unnamed 'friendly parties' have used Capella's data to track force build-up and to assess impacts on critical infrastructure. Koepke was hesitant to talk about other uses in Ukraine, but said in general Capella's data can be used for detecting aircraft and waterborne vessels, for spotting changes in the landscape and for humanitarian aid logistics.
Partnering for greater impact
After a time, Koepke's company was approached by data labeling company Scale AI, also in San Francisco, to partner with it in enhancing SAR datasets with AI and providing them for free. Scale was unable to provide an interview.
The Capella-Scale partnership turned out to be beneficial not only for Ukraine but for others as well.
'Part of being a force for good in this particular case, we were really happy to partner with them to ensure that we were providing these datasets in a more public manner through the commercial channels,' Koepke said.
The space tech company continues to position its satellites over Ukraine and provides SAR images to both paying customers and as open source.
'There are lots of entities that are doing analysis on our data,' he said. 'And sometimes they share that analysis back with us and other times they don't, but in all cases they're always asking for more. So, the data that we are providing for the Ukrainian effort is certainly in high demand.'
SATELLOGIC INC. 
