Chahab Nastar has always had a passion for patterns, whether he was tracking the popularity of his rock band Busy Being Born or designing software to distinguish unhealthy heart motions, which he did as a researcher at the University of Paris. Now he’s pouring that passion into helping computers understand everyday objects and scenes—a challenge that’s still “extremely difficult, if not impossible, for machines,” he says. Show a picture of your beach vacation to today’s average image-recognition program, and it can find similar beach scenes. But it can’t tell which beach is pictured or who is sunbathing. To move to that level, Nastar’s Paris-based startup, LookThatUp—now LTU Technologies—has enhanced image-recognition techniques with artificial-intelligence- based learning algorithms. Now, the more images of, say, cars and animals LTU’s system memorizes, the more quickly it can classify a Volkswagen or zebra. The system’s industry-leading recognition speed, a mere 200 milliseconds per image, has helped LTU sell software licenses to a dozen U.S. and European companies.