Will all the new computer vision stuff be coming to Sighthound Video?

March 23, 2017

One downside of having products for developers, enterprises and consumers is that sometimes we don’t do a good enough job of saying what each product can do. We’ve had people interested in Sighthound Video ask, ‘Where’s the tab for recognizing vehicles?’. In fact, just looking now at our homepage I see three press releases announcing the Sighthound AI Appliance, emotion recognition in Sighthound Cloud, and our new deeply learned license plate reader. All three are great, but they aren’t part of the same ecosystem. Not yet anyway.

Buried deep in the halls of Sighthound we have a team of PhDs. They aren’t allowed food or water, time off or weekends, just coffee and crack rocks. Every now and again they invent something world-beating. One of the latest things out of the lab was the car recognizer. It can recognize make, model and color with about 95% accuracy. There is a demo of it on the Cloud page of the Sighthound website. In fact on a long car journey last weekend the screaming crowd in the back seats were silenced for a while by bringing up the Sighthound Cloud page and taking snapshots of vehicles as we drove by, then seeing if Sighthound got it right. It beats 3 hours of I Spy.

At Sighthound we want to bring computer vision to everyone. We already have artificial intelligence in a consumer product that over half a million people use, which puts us in a fairly small universe. The more advanced computer vision in Sentry and the Cloud is targeted at our enterprise users. We have big companies putting our computer vision into their products and services, even though you’ll never know we’re in there. But these things tend to trickle down to the consumer market, and we’ll be no exception.