A small Russian startup company has one of the best facial-recognition algorithms in the world — and global powers are interested in getting a closer look.
Moscow’s NTechLab has created FindFace, a powerful app that has been downloaded in Russia 600,000 times since February. Its performance in Washington’s “MegaFace” competition in 2015 (it bested Google’s FaceNet), and in the real world, has sparked at least 300 inquiries from businesses and governments around the world.
NTechLab co-founders Alexander Kabakov and Artem Kukharenko recently agreed to an interview with Tech Insider to discuss their long-term plans, which are likely to worry privacy advocates.
“We want to cover all cameras across the world with facial recognition,” Mr. Kabakov, NTechLab’s marketing chief, told the website on Tuesday. “We see that we are leaders in this sphere.”
Mr. Kukharenko said critics’ claims are unfounded since the company’s software scavenges the internet for publicly available information.
“We work like a search engine, like Google,” he said.
“FindFace is an instrument. So of course people with bad intentions can use it,” Mr. Kabakov added.
Tech Insider noted that a photographer spotlighted the power of FindFace in a recent series titled “Your Face Is Big Data.”
Egor Tsvetk, 21, took pictures of random strangers in a St. Petersburg subway and then used the technology to identify them. Those individuals then found out their identities were revealed via media coverage of the project.
“As a person, I certainly felt some shame,” Mr. Tsvetk told the website. “But in this project, I tried on the role of internet stalker.”
Mr. Kabakov and Mr. Kukharenko said that security firms and governments have similar technology anyway, but NTechLab’s is public. They also note that FindFace recently identified two men suspected of arson in St. Petersburg.
“We see that the advantages for society from our technology are more helpful [than harmful],” Mr. Kabakov said.”
• Douglas Ernst can be reached at dernst@washingtontimes.com.
Please read our comment policy before commenting.