This article first appeared in Unlisted & Unlimited, The Edge Malaysia Weekly, on June 13 - 19, 2016.
IMAGINE a digital billboard that looks back at you while you are looking at it, taking in your age, gender, ethnicity and emotional reaction. How much more effective would marketing campaigns be if companies knew exactly how people reacted to them and how this translated into the mighty dollar?
This is the technology that Glueck Technologies Sdn Bhd has come up with. The company creates products using a combination of artificial intelligence, deep machine learning and computer vision.
CEO Alberrt Alexander says the technology turns any camera into a human-like visual intelligence machine. “We have filed a patent application for this technology and are waiting for it to be granted,” he adds.
He says the company mounts cameras on digital signboards and picks up information on the people who pass in front of them. “We identify their age group (child, young adult, adult or senior), gender and emotion. How do they react to the ad? Are they sad, happy, surprised, angry, content, disgusted, or are they neutral?”
The technology also measures their attention span — how long people look at the billboard or whether they look at all — as well as the ad’s hit rate. “Say 1,000 people go past the digital screen and only 300 look at it — that means the conversion ratio is 30%,” says Alexander.
Glueck, which means happy in German, was established in 2014. Today, it operates with just seven staff.
Co-founder, inventor and chief technology officer Imran Akthar had worked with co-founder and COO Theenathayalan Parthasarathy at Matrix View Technologies in Chennai, India. This company was eventually listed on the Australian Securities Exchange (ASX).
After leaving Matrix View, they went their separate ways — Imran to Finland, where he worked for a company that created products for Nokia, and Parthasarathy to Canada, where he worked for TUC Inc, doing banking solutions.
Imran had a background in artificial intelligence, deep machine learning and computer vision. He says, “When I moved back to India from Finland in 2010, I had an idea about doing something in the advertisement space. The idea was about understanding human demographics such as age, gender, ethnicity and emotions. What if the machine knows the audience in front of it?”
He reunited with Parthasarathy, who had just returned from Canada, and the duo worked on this idea. Within two years, they had a working proof of concept.
“At this point of time, our technology was able to recognise the audience’s age, gender and basic emotions such as whether a person is neutral, happy or surprised. Using the technology, we were enabling our cameras to have human-like machine intelligence — what we see with our eyes, trying to translate that information to a machine,” says Imran.
There were a few issues here such as privacy. How could this information be translated so that it provided important feedback to media owners while not compromising on the privacy of the viewers?
Simple. The images are not recorded. They are translated into the relevant data and fed to the cloud, where the information is run through analytics and turned into relevant feedback for media owners or advertisers.
Up to this point, the project had been funded out of their own savings. They had put the equivalent of RM100,000 into the development. But they also knew that they would need outside help to take the product to the world stage.
This was where Alexander came in. He had been a project director at RHB Bank and Parthasarathy had worked for him in the capacity of a vendor. The duo knew he had more than 30 years of experience in the banking industry and the right connections.
“We flew to Malaysia and met up with Alberrt, and things just clicked. The three of us shared a common dream of taking the company forward,” says Parthasarathy.
Alexander nods. “In my former life, I was in banking and IT. I was developing an application on a document-imaging solution for the bank. Pat [Parthasarathy] and Imran used to work for a company in Chennai, where they compressed data and push it out on a very small pipe. We shortlisted their company for the job.”
The founders knew Malaysia was a great place for start-ups, with a lot of government support and funding. So when Imran and Parthasarathy popped the question, Alexander said yes.
Parthasarathy says, “There was opportunity, government support, MSC status, early-stage funding. We found a few Malaysian investors. We filed our patent in Malaysia, India and the US, and the Patent Cooperation Treaty gives us 30 months to file in 145 countries.”
The company also received RM500,000 from the Cradle Investment Programme (CIP) 500 to commercialise its product. It had four staff at the time, then it brought in five advisers, including a psychologist and an advertising veteran.
“Dr Jay Brinker is a lecturer and researcher from the Swinburne University of Technology in Melbourne, and she specialises in neuro recognition and emotions. Spencer Wing is the co-founder of advertising company Spencer Azizul Sdn Bhd, and he has over 30 years of experience in that industry,” says Alexander.
“Brinker’s major research areas are neuro recognition and emotion. She explores the experience of emotions and thoughts, and the neuro function that relates to and underlies them. She is also a registered clinical psychologist.”
Once they had secured the initial round of funding from friends and family (of about RM400,000), they started Glueck Technologies, opened their office in Malaysia, hired resources and identified a gap in the out-of-home (OOH) advertising market for this product. Then, they started pitching their product to media owners and agencies.
Alexander says funding has not been an issue for the company. “A lot of people want to invest in our technology. We have had no problem getting funds.”
Perhaps this is because the company is operating in a “sexy” area. After years of being a non-starter, artificial intelligence is coming of age and the big boys are starting to take notice. For instance, in January, Apple acquired Emotient, which uses artificial intelligence to detect emotions and analyse facial expressions, and late last year, Amazon.com acquired Orbeus Inc, an artificial intelligence company that focuses on image processing.
“Emotient is a market research company and Apple is trying to find out exactly what they want to use it for. They like the technology and they might as well grab it before somebody else does. Everybody has a ‘kiasu’ attitude when it comes to this,” says Alexander.
According to market research company ReportLinker, the emotion recognition and detection market is projected to grow to US$22.65 billion by 2020 from US$5.66 billion in 2015, a compound annual growth rate of 31.9%. The other major players in this market are Affectiva, Kairos AR, nViso, RealEyes, SightCorp, Noldus Information Technology, Sentiance, Tobbi and IBM Watson Foundation.
At the moment, respondents can answer questions about an advertisement they are shown by rating it on various scales, according to Terry Lawlor, executive vice-president of product management at Confirmit, which provides customer experience software solutions.
“While broadly effective, it relies on the respondent’s ability to recall what they have just been shown and how they interpret their emotions as well as their ability to express these emotions in the way the researchers need the information to be gathered,” he says.
Researchers can also observe and record emotions while video content is being shown. But this needs specific skills and is difficult to perform consistently.
Lawlor points out in an article on Populus, which provides business and consumer insights, that technology which monitors facial expressions bypasses these issues by capturing data as the respondent views a video. This type of technology builds a huge reference database of expressions against which to judge the face being viewed. Researchers can then compare the aggregate emotional performance of their video clip against a benchmark.
So what makes Glueck different from the other companies doing the same thing out there? “We are the only company that provides emotion-based audience measurement on hardware. Also, our technology can be used by developers by using Cloud API. What this means is that another company can build an application from scratch using our technology,” says Imran.
For example, one of its customers uses its technology to measure the number of people in a shopping mall using Wifi and Beacon. “They have integrated our technology into their product to make the data more reliable and accurate,” he says.
Basically, this technology fills a gap in the OOH market. “There was a report by PricewaterhouseCoopers that said there was a big gap in the measurement for the OOH digital market. People just put up signage but have no idea how effective it is. Our first product — the A3 — is sort of a Google analytics for digital billboards,” says Imran.
“Media owners spend millions of dollars buying digital signage and assuming that installing them in crowded areas, such as malls or transport hubs, will give them more eyeballs. Our product verifies and authenticates this information, which enables the media owners to price their inventory accordingly.”
Glueck launched the A3 at the InnoDesignTech Expo in Hong Kong last December. It was looking for “reference” customers to test out the technology. It secured one from Thailand at the expo itself, and others from Malaysia, Singapore, North America and Algeria as the word slowly got out. “We got a good response from the medical industry and security, retail and OOH markets,” says Imran.
During the initial rollout of the projects in these countries, they were able to provide valuable feedback to media owners on how effective their billboards actually were and how to price them accordingly. For instance, one company, which had invested a great deal of money putting up digital signage in a train station, found that it was largely ineffective as people were rushing on and off trains and did not have time to stop and view the advertisements running on the billboards. The conversion rate was a pitiful 9%. Another company found that the billboards put up in shopping centres, where people moved at a much more leisurely pace, were pretty effective. These had a 40% conversion rate.
What are the medical applications? “We had a request from Australia to use this technology to gauge the level of pain a patient is going through,” says Imran.
At the moment, technologies to measure pain are either ineffective or intrusive, for instance, asking a patient to rate his pain on a scale of 1 to 10, or putting electrodes on various parts of his body for measurement purposes. With the new technology, the pain can be gauged from the expression on a person’s face.
In terms of security, the applications are twofold. On the one hand, an immigration officer will be able to run the person’s face through its database to see if it matches any known threat. On the other, they could also measure things like whether the person is sweating or looking fearful, which would send up a red flag.
And then there’s the retail sector. “Our customers/distributors in Singapore are using our technology to understand the customer experience in high-end shops. For instance, if you have launched a new product or product line, you want to understand how your customers react to it,” says Imran.
“There is one more industry that could use our technology — animation,” he adds. At the moment, the most advanced animation uses actual people to monitor the changes in facial expressions or movements. They put physical markers on their faces and this information has to be translated into a 3D avatar.
“What we are doing is eliminating the first step. The camera captures 256 points on the face and this can be translated into the 3D avatar without the need for physical markers. The process is less time consuming and more cost-effective,” says Imran.
Glueck will be marketing this to animation studios in Malaysia, Singapore and other countries in the Asia-Pacific. It has not set its sights yet on any of the major animation studios in Hollywood or Japan, but as Imran quips, “If they approach us, we will sell it to them.”
He admits that this last application has the potential to be one of the most lucrative applications for this technology.
The team has been hitting the exhibition/conference circuit pretty hard. Earlier this month, it participated in 57th FEPE International Congress in Barcelona, Spain, where it launched the second iteration of the product, A2, which enables cameras to identify the audience watching the billboard and select which advertisements to run based on their demographics.
“We are super excited about this product. It is going to change the whole industry and we believe it will be a multibillion-dollar industry. We are going to share details with our partners in Malaysia, Singapore, Thailand, the US, Algeria and Hong Kong,” says Imran.
Next month, Parthasarathy says, the company will participate in the Tech in Asia conference in Bangalore, India. “The reason we are participating in all these exhibitions is to showcase our product line and explore new markets as well as to look for potential strategic investors,” he adds.
This is a potentially explosive industry, but the company is being very conservative with its income projections. Parthasarathy says it expects to make RM400,000 this year and US$2 million next year. It is only projecting to break even by 2019.
It hopes to be listed within a year, possibly on the ASX because it is the most welcoming stock exhcange for start-ups. As long as the technology is good and has potential, it does not matter if the company is not making money yet. It does not have to show a track record. The ASX considers future potential rather than past performance.
“There are complex procedures out there for a start-up to raise funds in any stock exchange. Australia has made it very easy for a start-up to list. The company we worked with [Matrix Views] raised over A$300 million by listing on the ASX,” says Parthasarathy.