AI Keeps IDF Satellite Imagery Up To Date By Mixing Old And New

Advanced artificial intelligence software is helping Israeli forces currently operating in the Gaza Strip strike their targets with the utmost precision, by accurately mapping ground-level data over real-time images and videos taken by drones and satellites. 

The software, developed by Israeli startup Edgybees, aligns aerial video and satellite imagery over precise reference photos of the area that have since become dated, making it easier for the Israel Defense Forces (IDF) to pinpoint specific locations and send in ground troops to that exact position. 

It does so by analyzing all of the common features between both the old and new aerial photos – from seemingly minor landmarks such as a fork in the road to more prominent objects such as rooftop solar panels – and uses them as anchors to rotate and align the new data until it perfectly overlays the original reference material. 

Menashe Haskin, co-founder and chief experience officer of Edgybees, tells NoCamels that ensuring such precision is crucial, and in many cases can even be a matter of life or death. 

“What most people don’t realize is that when you get a satellite image that includes ground data, it is never accurate,” he explains. 

“It could be several hundred meters off [target]. And when you’re trying to derive operational value from it, accuracy becomes immensely important.”

Incoming satellite and drone imagery, Haskin says, must be layered over existing reference images of the same area, so that intelligence, defense or search and rescue units can determine the precise location they want to reach.

But it’s not as easy as simply dragging and dropping the new imagery over existing reference data, Haskin explains. 

The satellites used to snap these pictures are continuously orbiting Earth from outer space, so the perspective from which each photo is taken can and will vary, making it difficult to manually align the older and newer images. 

In fact, says Haskin, it can take a professional hours to properly align a single photo by hand. 

Different lighting conditions, increased cloud cover and other weather conditions may also affect the ability to make out all of the features in a new image, adding to the difficulty of accurately aligning one image over the other. 

“But with our technology, you can do this automatically,” says Haskin. “Within a minute, you have the alignment and can see the differences.”

Other image-aligning tools exist, but according to Haskin, they aren’t as accurate as Edgybees’ platform. And unlike its competitors, he says, his startup’s technology can even be used on thermal imagery, which can be helpful for defense purposes and operations. 

The somewhat unusual name comes from an encounter one of the co-founders had with a swarm of bees while testing an early version of the technology.

“Some people love the name and some people hate it, but it’s so unusual that nobody forgets it, and that’s a good thing for a startup,” Haskin jokes. 

The IDF is not only using Edgybees to increase the accuracy of the targets it is striking, but also to enhance and improve communication between troops. 

Once a video has been aligned correctly to a reference image, soldiers can zoom in or out, add graphics and even mark specific points on it. And as the video plays and the images move, the added information moves with them.

This, according to Haskin, allows someone to communicate a specific location to another person by referring to the added marks on the screen instead of trying to describe the spot. 

“Without Edgybees, everything shakes,” he says, referring to videos whose added marks are not seamlessly coordinated with the footage. “You put down the marker and cannot even know if it’s in place.”

Defense applications aside, Edgybees’ software has also been used by government agencies to monitor mining operations, insurance companies to review hurricane and tornado damage, and even by planners creating power grids. 

Edgybees, founded in Herzliya in 2016 and today headquartered in Maryland, was initially created to design augmented reality games for amateur drone users but the company soon realized that the tech they were developing had more consequential potential. 

“We realized that there was a real need for this kind of technology, and that there was a large gap between what people were looking for and what was available back in the day,” Haskin recalls. 

The software was first used by emergency response teams during Hurricane Irma in 2017 to locate people who were stranded near washed out bridges in Florida. 

Since then it has been used countless times by emergency services, from aiding 15,000 firefighters battling wildfires in California in 2017, to helping rescue efforts following the damage of Hurricane Michael in 2018. 

“This is a very, very powerful tool that we have brought to the table,” Haskin says.

Subscribe to NoCamels weekly newsletter and get our top stories