Faulty utility equipment ignited more than 2,000 fires in California over a three-year period and power lines are the third most common cause of fires in the Golden State. Buzz Solutions Inc. is tackling the problem with machine learning.

The three-year-old startup, which that was born out of the StartX incubator at Stanford University, Friday said it has raised a $1.2 million seed round to build and market a technology that uses image recognition to spot faults in utility equipment before they become dangerous.

This platform analyzes hundreds of thousands of images taken by drones, helicopters and fixed-wing aircraft find equipment flaws and other fire hazards like overgrown vegetation. The company says it can reduce the time required to do a comprehensive analysis of a utility’s field assets from months to hours at a fraction of the cost.

Drones have already been a godsend to the utility industry. Navigant Consulting Inc., a subsidiary of Guidehouse LLP, has estimated that they can cut the cost of a single transmission tower inspection by 96% while improving employee safety.

The problem is that once images are captured, “the process on the back end is manual,” said Kaitlyn Albertoli, Buzz Solutions’ co-founder and chief executive. “Utilities have linemen and engineers who map images to a certain structure to look for failure modes.  It works well with a small number of images but it’s not scalable.” It can take six to eight months for a utility to crunch several hundred thousand images manually, by which time some of the results may no longer be relevant, she said.

Buzz Solutions’ machine learning technology is trained to look specifically for aberrations that indicate a possible fault. The founders claim it can detect three to four times as many faults as competitive products with an accuracy rate of better than 95%. The company is in trials with utilities it declined to name in the midwestern U.S., New York and southern California.

The image recognition technology can be tuned to the needs of the utility, said Vikhyat Chaudhry, co-founder and chief technology officer. “For example. in southern California it’s mainly vegetation that becomes a fire hazard, while in the midwest they’re more concerned about corrosion and rust,” he said. “We’re using multiple fault signatures depending on the geography.”

The six-person company has bootstrapped its operations to this point, raising funds through incubators and grants. Albertoli said the plan is to convert some pilot projects to customer licenses prior to raising a Series A round.

“The great thing about utilities is that once you get into their vendor list the turnover is low and they tend to recommend you,” she said. The company also intends to offer the platform as a service to independent drone operators on a subscription basis.

The funding round was led by Blackhorn Ventures Capital Management LLC with participation from a syndicate comprising Ulu Ventures LLC, Vodia Ventures LLC, and Advisors.fund LLC.

Photo: Unsplash