BU Alumni Target Advertising, Journalism With Visual Recognition Technology

in Alumni, Awards, Faculty, Graduate Programs, Graduate Students, News-CE, News-ISS, Recognition, Research-ISS, Students
September 11th, 2012

Screen Shot 2012-09-05 at 9.39.06 PM

On their website, ReKognition.com, Orbeus offers visual recognition technology that identifies facial expressions.

Tagging your friends on Facebook became a lot easier when face recognition technology came into play and identified familiar faces from your previous photos.

Orbeus, a joint BU-MIT alumni startup, saw the potential of using visual recognition in other fields and decided to target new audiences by molding their services for advertising agencies, social media networking sites, and journalists. The technology they’re developing can identify brands, logos, and other information that can be helpful for advertisers to find the right target audience.

“Visual recognition is in demand and people in surveillance, e-commerce, and advertising are already interested,” said Xing Meng, one of the company’s co-founders. Meng Wang (MS ’12) and Tianqiang Liu (MS ’11) also make up the company’s founding entrepreneurs.

The idea for Orbeus first developed in Boston, where Meng was a student at MIT and Wang and Liu were studying at BU.

“I had been studying lots of topics such as anomaly detection, image saliency, and 3D reconstruction,” explained Wang. “From that, I knew that I wanted to help develop a unifying data-driven framework that could tackle a lot of computer vision problems, which later on became the technology foundation of Orbeus.”

At BU, Wang said he enjoyed freedom in his research and the academically dynamic community.

“I love to play with different ideas and patterns, but it was only through in-depth discussions with my MS professors that those ideas eventually transformed into concrete algorithms that are grounded in sound theory,” said Wang.

Professor Janusz Konrad (ECE) was one of the professors who helped shape Wang’s ideas.

“In his research projects at BU, Meng [Wang] showed an exceptional ability to solve practical image processing problems with great algorithmic efficiency,” said Konrad. “This often required integration of several algorithms into a complete working system, so it comes as no surprise to me that he translated these skills into a successful startup.”

Wang’s creativity and hard work came to fruition at the MIT $100k Entrepreneurship Competition, where team Orbeus was a semi-finalist. From that contest, Orbeus gained more than a formal mentorship, an expense account, and the opportunity to pitch to a group of judges in a simulated funding meeting. The mandatory demo presentations and business plan write-ups forced the members of Orbeus to make concrete decisions.

“By asking us to present demos, the competition made us shift our focus from technology-development to product-building ,” said Meng. “It forced us to sit down as a team and decide where we wanted to take Orbeus as a company.”

Professor Prakash Ishwar (ECE) wasn’t surprised to hear about the success Orbeus found at the MIT competition. Like Konrad, he also had the chance to work with Wang at BU.

“Meng Wang is an ambitious, fearless, and gifted hacker who has boundless creative enthusiasm for tackling real-world visual information processing problems,” said Ishwar.

Ishwar added that his former student showed potential even before joining BU. As an undergraduate, Wang launched GoGoStudio, a blogspot that documents his handiwork in machine vision.

Like any new tech startup, Orbeus has faced its share of challenges. After initially developing an infrastructure without picking any direction, the group’s many mentors began pulling them down different paths.

“With at least six meetings a day, our team had to sit together and evaluate what we needed to do from a technology perspective,” said Meng.

After sorting through their different options, they decided on one clear direction for marketing in terms of the right target audience and are working on developing their product with that in mind.

The second challenge was funding. Paying for experienced developers, hosting, software, and hardware added up when the company expanded. Luckily, members of Orbeus were able to secure additional funding when they were introduced to others who needed the technology their company was offering.

In May, the company moved to Excelerate Labs. The Chicago-based incubator gives the team mentorship, products, business, marketing and a platform to pitch and showcase their product in front of 400 to 500 people during a demo day. Meng believes this could potentially help them secure their second round of funding.

Moving to Chicago also gave the team the opportunity to find more advertisers that were interested in the product. In the past couple of months new opportunities have opened up. In June, Facebook acquired Face.com, a leading face recognition service company with a third of the visual recognition marketplace, opening up a huge hole for other startups like Orbeus to fill.

The technology has the potential to benefit journalists as well. Current stock image websites limit searches to generic pictures, but with visual recognition, it will be possible to search for specific attributes of images such as negative emotions in facial expressions.

With new opportunities arising, the company has big goals for the upcoming year. During the first month of August, they released a new Application Programming Interface (API) allowing developers to test and sign up for an API key to try the platform on their website, ReKognition.com.

Two weeks later with 200 developer sign ups, they launched their first site for consumer search images which will allow customers to find photos with themselves and their friends in an easy-to-use platform.

Alongside these new projects, the team hopes to increase user adoption and find more developers in the months to come.

-Sneha Dasgupta (COM ’13)