Video analytics platform Valossa just launched Val.ai, a platform to help video creators, advertisers and other video boffins figure out what’s going on in video. In addition to computer-vision tricks (“Man on a beach”, “car interior”, “kitten is surprised“), the platform can do sentiment analyses (person is happy / person is sad / person is confused) and even heart rate analysis based on a high-definition video stream alone.
I tried the platform out at TechCrunch Disrupt and was bowled over by how much the company knows about me, based purely on a video feed. The heartbeat feature, especially, is a neat trick. The company uses microscopic changes in the image feed to detect when heartbeats are occurring – and if my Apple watch’s heartrate monitor is anything to go by (it might not be) it’s actually remarkably accurate. Even if it proves not to be a hundred percent precise on an absolute basis, it could be useful for people analyzing video footage to see relative (i.e. heart rate increasing / heart beat slowing) changes.
“There are many uses for our technology,” explains Ville Hulkko, the company’s chief commercial officer. “Especially for people working on huge film projects. Imagine you are looking for a particular piece of footage of a dog and a ball on the beach, for example. If you don’t remember when it was taken, you’ll spend a long time looking for the correct video clip. With our technology, it’s easy: You just search for it.”
The company’s sentiment analysis engine is eerily good. You can try it out on the What is my Movie website, launched at Disrupt last year — try entering “dog gets lost” or “hero wears a suit” or “money is burned”, for example. Awesome, no?