SINGAPORE (Oct 28): Visual artificial intelligence company Prisma AI began life as a business intelligence unit until 2003, when it started a project with Fraunhofer Institute to scan and piece together shredded letters between the former East Germany and West Germany.
The shredded letters were placed on a conveyer belt with cameras mounted on top, after which Prisma’s algorithm was used to reconstruct them digitally, according to Shreeram Iyer, chairman and group CEO of Prisma AI. “That was the first project. We’ve done 2,000 bags [of letters] in 12 years out of 16,000 bags,” he says in an interview with The Edge Singapore.
That first project led to another one with Interpol to prevent stolen art from being sold, says Iyer. Mid-tier auction houses often do not list owner information, making it easier for stolen art to be auctioned off. Interpol agents utilise Prisma’s algorithm by taking pictures of the paintings put up for auction to compare them with pictures of stolen paintings stored on a server. “If there is a match, they can stop the auction and do carbon dating and analysis [to verify the painting],” he explains.
Google, which has numerous online applications that are free, unveiled its image recognition capabilities as part of its broader search offerings. Prisma differs in that it specialises in this field of image recognition and is purely focused on building a business model out of helping businesses solve problems with image recognition and analysis. Prisma is also one of the few companies with their own image recognition engine and crawler to search for and reference images.
The company has sold its services to big-name customers. For example, BMW is using Prisma’s solutions for parts recognition, according to Iyer. Parts that fell to the floor during manufacturing would often be vacuumed up or swept away, costing the German carmaker about €14 million ($21.1 million) a year. Now, parts can be recognised with an app and a camera, and directions given to employees on where to return the parts.
“They [also] use it for training — if I’m an engineer, I need to know how this [part] fits into my car. I just use my phone and take a photo, [and the app] will show an animation of how it fits into the car,” says Iyer.
The company is moving from projects to providing software as a service (SAAS). Prisma has solutions that can recognise unattended baggage or unidentified objects in airports or railway stations where surveillance cameras are placed.
“What [the solution] does is, when it finds an unidentified object, it will automatically trigger a push notification to a security [guard] nearby. The security guard has to open the application on his phone, and scroll through the video recordings to find a match for the person who left the object behind,” says Iyer.
The solution will allow security to see who brought the object, where they were coming from and where they went after they left it behind. It is also used to recognise number plates on cars and can help the authorities debit outstanding fines from the vehicle owner. “We can use any camera; we just give the software to the camera owner or the servers and it is integrated automatically. This gives the camera a brain, and it is no longer a dumb piece of hardware,” Iyer explains.
Prisma can also do sentiment and emotion analysis from camera footage. The algorithm analyses each frame captured by the camera against a reference provided and the difference will indicate the person’s emotion and sentiment, says Iyer.
The company’s solutions can help in the understanding of behavioural patterns too. For example, Prisma’s technology can detect from a person’s posture whether he or she is carrying a bag heavier than usual.
“There are a couple of AI-based applications for the camera. The camera becomes intelligent and makes decisions based on you and the pattern of understanding the data. We are camera-agnostic; we started from [low-resolution] megapixel cameras to HD cameras,” Iyer says.
Aside from moving towards monestising its technology via an SAAS model, Prisma is also looking at a potential listing on the Singapore Exchange. As part of the listing, Prisma moved its headquarters from Germany to Singapore in November 2017. Iyer says the company is looking to do either an IPO or a reverse takeover (RTO) of an SGX-listed company.
“[Listing] will give us a platform for the global market; it gives us credibility as a market player,” he says, noting that Prisma’s competitors have already gone public. Part of the reason is that governments and organisations prefer to do business with a public-listed company.
“We can get [private equity] funds any time, but a listco gives you credibility with the customer. When we say we are a Singapore listed company, to customers and partners, it is a landmark,” says Iyer. “Singapore is known as a financial hub in Asia; the structure is very clean here. Once you are listed in Singapore and known as one of the leading players, it will be easy for us to enter any market worldwide.”
Prisma’s plan to go public has since taken a slightly different turn. On Oct 3, it entered into a joint venture with SGX-listed management consulting firm Forise International.
Prisma will hold a 49.9% stake in the JV company, Prisma Technologies, while Forise will hold a 50.1% stake. The new company will provide corporate and financial solutions, advisory services and management consulting services in the IT field of visual security.
The JV will be a subsidiary of Forise, and will manage Forise’s visual-based contracts for airport, metro railways, highways and infrastructure projects. Prisma will be paid $4 million for working capital and loss of profit. Prisma Technologies won a US$2.34 million ($3.2 million) contract on Oct 21 to provide visual-based security for aviation and highway automation systems in Africa and Asia.
