A normal video surveillance camera mounted on some building can easily confuse a burglar with a stray dog and send false alerts to the security office. That’s because it can’t always read shapes in real time, just detect motion. Fatigued by false alarms, security guards may eventually tune out the warnings. Umbo Computer Vision, a Taiwan startup, figured out how to get a better read from those cameras, cutting back false alarms by 20 times, a co-founder says.
Venture capitalists believe it. They have put a combined $10 million into the 3-year-old company that began selling its alarm systems in September 2016. Umbo has sold 200 systems, including to a Taiwanese university with 10,000 students but mostly in the United States and some to Fortune 500 corporate clients. That amount of capital is high for a Taiwanese startup, the head of Umbo’s accelerator program says.
Here’s how it works: The firm’s staff uses artificial intelligence to program cameras so they read shapes in dark, stormy conditions. That process eliminates dogs and cats from the stream. Real-time video from the cameras goes into a cloud for security. Hackers might otherwise get into the cameras themselves. Artificial intelligence also records images of real security threats as a learning point for other cameras.
Umbo and others had been trying to rig up something more secure and less error-ridden over the past five years, says Shawn Guan, one of four co-founders. Total investment in artificial intelligence, or AI, last year came to between $8 billion and $12 billion, according to the McKinsey Global Institute study.
Guan, 30, “stumbled” into security more than a decade ago after studying biology and computer science at the University of California, Santa Barbara. He was working for another security startup as the guy who talked with clients. He racked up 2,000 meetings or calls over that period. “I went to a lot of sites, just to go there and see what happened,” he says “Just too many false alerts.” That company bloomed into an initial public offering.
He and fellow co-founders, all Taiwanese, have built Umbo staff to 40 people, 10 with artificial intelligence skills. It sells hardware for around $400 and software for $300 to $600 depending on the number of streams. The firm contacts for the manufacturing of cameras.
The Taipei-based startup found funding based partly on advice from an accelerator program three years ago. That program also opened channels to suppliers and talent for the firm. Venture capitalists normally help only artificial intelligence startups run by professionals who can show their technology will bring dramatic changes, says Jamie Lin, founding partner of the accelerator, AppWorks Ventures in Taipei. They’re looking for “a disruptor” that “fundamentally changes how things are done, for the better,” Lin says.
Umbo is “still spending” the venture capital rather than profiting on sales, Guan says, but it’s aiming for its own IPO by 2022.
Original story at Forbes.