On the starting block

Asaphus Vision GmbH has developed innovative new facial recognition software that enables highly complex calculations – in a resource-efficient way without an operating system

Dr. Lenka Ivantysynova © Asaphus Vision GmbH

The eagle is the king of the sky. It is a sublime and powerful creature but, alas, much too obvious. “When it comes to excellent vision, everybody immediately thinks of an eagle”, argues 37-year-old Lenka Ivantysynova. But who would ever think of a trilobite? And, more specifically, who would think of Asaphus kowalewskii, an arthropod and sea dweller that lived roughly 450 million years ago. This animal’s unique feature was it protruding and independently moving pair of eyes. “We wanted a name that entailed extraordinary vision in the animal kingdom. We did our research and came across the Asaphus”, remembers Ivantysynova, who also has a fossilized copy of the trilobite on her shelf at home.

It’s a hot July day and Ivantysynova is sitting in her office at the Charlottenburg Innovation Centre CHIC. “Sitting” is perhaps the wrong verb, as Ivantysynova and her nine colleagues would be better described as “standing” on the starting block. Indeed, she and her clients are working hard to ensure that her software gets integrated into serial automobile production. “As far as we’re concerned, our software is more than ready for the market”, says the computer scientist, shrugging her shoulders and laughing. She knows it’s not always so simple. “Car manufacturers are huge corporations that do everything according to strict rules”, says Ivantysynova. The slow pace of business is one thing, but the automotive industry is also one of the most safety-conscious sectors in the world. The systems they use must be utterly dependable, especially when it comes to sensitive technology such as driver assistance systems.

Asaphus has developed innovative new facial identification software for use in exactly such systems. The software uses a small camera to detect the direction in which the person at the wheel is looking, as well as which direction their head is facing and whether their eyelids are open or closed. This is important for detecting fatigue. The software also enables rapid facial recognition; indeed, it can recognise a face within milliseconds. “Of course, faces change over time. This is why the software automatically performs its own scans”, says Ivantysynova, who founded the company together with four fellow students from HU Berlin.

Facial recognition software might sound quite simple in theory. However, in practice, it requires special algorithms and a high-performance computer. As Ivantysynova reveals, “Not only are the calculations highly complex, they also have to operate extremely quickly”. And this is exactly what Asaphus does. Plus, it does so entirely without a high-performance computer. In fact, this software doesn’t need any operating system; it’s compatible with all hardware and has its own memory storage management. In addition, the programme consumes very few resources.

“In other words, we generate very little heat, which is an important issue in any vehicle”, explains Ivantysynova. Indeed, nobody has been able to achieve the same result so far. 

As already mentioned, sitting is the wrong verb to describe Asaphus. The team is already waiting on the starting block.

 

Stefanie Paul
Kluge Köpfe 2017