TwentyBN, an AI startup with offices in Berlin and Toronto, has announced that it just brought in $10 million through a funding round led by M12, Microsoft’s venture capital fund.
As VentureBeat reports, the startup is focused on computer vision technology designed to interpret human actions. TwentyBN has been training its AI technology on a database of about two million crowdsourced video clips depicting a wide array of human behaviors, from walking up a staircase to drinking from a glass of water.
The aim is to endow AI with a general understanding of human behavior, with an MIT research project having recently used TwentyBN’s datasets to create an AI model that could predict what humans were going to do with certain objects, such as ripping up a sheet of paper. Other researchers have used the datasets to train a range of gesture recognition AI systems, and it’s easy to imagine this technology being applied in the field of behavioral biometrics security and even physical security, with AI systems able to assess the intentions of individuals on surveillance camera feeds, for example.
Discussing the technology’s potential, M12 managing director Samir Kumar explained that TwentyBN’s AI models “recognize both the nouns and verbs describing a visual scene,” and so “[n]ew intelligent camera experiences become possible when these models derived from their crowd-acting datasets are deployed to run efficiently on IoT edge devices.”
M12’s investment was delivered along with funding from Coparion, Creative Edge, and MFV Partners, and TwentyBN says it will use the capital infusion “to scale its business,” according to VentureBeat’s report.