With so many companies vying for market share, how does the customer know what to select?
The advent of artificial intelligence (AI) has paved way for a new arms race in the world of technology. Several companies across the globe are now fighting to come up with algorithms that are as intelligent as humans and as efficient as machines. Note that this includes the likes of tech giants such as Google and Facebook. The competition is tough, but the stakes are high and possibilities, endless.
But with so many companies vying for market share, how does the customer know what to select? An obvious answer for many is to go with a well-known brand, say for example Google. After all, the search-engine maker has excelled at many things, so you can't go too wrong with it…right?
Well, not necessarily. Just because a company is good in a particular field doesn’t mean that it can replicate its success in an entirely new segment. There are countless examples of large companies that have tried entering new markets and failed miserably. Selecting a product based on its maker’s expertise in a different field would be like shooting in the dark.
But then, what options do the customers have? It’s almost impossible for an average customer to test each and every solution out there.
This is where media reviews and more importantly, contests run by third parties, become crucial. One such contest is held by the University of Washington.
Megaface Challenge, as it is known, sets up global face recognition algorithms to face off against each other. The contest has brought to light some of the most efficient face recognition solutions out there. Google’s Facenet did quite well, becoming one of the top performers.
However, in 2015, a Russian startup that was founded in the same year beat Google and about 90 other teams to become a winner. Called
Ntechlab, this company is focused on applying deep learning technology to solve challenges relevant to several industries ranging from customs to security and law enforcement.
The Russian rival to the likes of Google and Facebook
Ntechlab’s facial recognition algorithm is available on FindFace.Pro cloud as SaaS and via SDK to any business that needs such a technology. Speaking to asmag.com, Mikhail Ivanov, CEO of Ntechlab, explained the technology behind the solution.
“We have found a special type of internal architecture for neural networks, that perfectly fits the face recognition tasks,” Ivanov said. “We use 2-phased search, the first phase is extremely quick and the second one significantly increases the accuracy. Due to that our search index consists of just numbers and the overall search time is about half-a-second. Up to this moment, our algorithm has performed around one quadrillion of photo comparisons, so we used that big data to raise the efficiency of our algorithm even further. Since Megaface we managed to significantly increase our performance further.”
The algorithm is based on the neural network that is capable of learning and determining distinguishing features for personal identification – eye size, eyebrow thickness, lip shape and so on. It was trained with millions of mapped photographs of people, in semi-automatic mode where the people in the photos are identified by specific names. Then the network learns by itself, trying to extract the vectors of features that would help solve the task. It determines attributes, assigns them certain importance and builds interconnections between them, and as a result, the network generates about 160 numbers to describe the information about a face. Having “calculated” the features, the neural network can apply them to other photos as well.
Security and other factors to drive demand
Ivanov believes that the company has the best technology at the backend which allows its customers to apply it to various business scenarios. He identified several factors that are expected to drive demand for Ntechlab’s solution.
“New types of the algorithms based on neural networks can be used in identification scenarios with video surveillance in public safety and commercial applications like in retail industry,” Ivanov said. “People already believe that facial recognition works and are waiting for real working solutions for different verticals. AI-based solutions have proved their efficiency in comparing with the people in some routine processes and people understand how they can earn from it.”
Ivanov added that recently, the company solution’s efficiency was confirmed by NIST, a measurement standard lab that has released a report evaluating leading facial recognition technologies that may have a wide range of civil, law enforcement and homeland security applications.