September 01, 2015      

To master Deep Learning, it takes a good understanding of both neuroscience and computer science. At ViDi, we have the right combination of both. Based on the latest understanding of the well hidden tricks nature uses to process and reason on visual stimuli, a novel way to analyze images actively has been developed, resulting in a product that performs like humans, is self-learning and easy to implement.

In one example, a customer spent 60 days of hard work to develop the machine vision part of a robotic application able to sort real wood pieces from MDF, reaching an accuracy of about 80%. He was going to lose the project because 80% was not good enough. With ViDi suite, it took only 15 minutes to get to 98.5% accuracy. In another example, a famous Swiss watchmaker was tired of manually inspecting decorated watch parts. He was using upto 60 operators to perform that specific task, with questionable accuracy. With ViDi suite, he was able to replace all these operators with only 5 autonomous robotic cells. And we do it with our generic tools, No programming, No optimization, On a single GPU.

ViDi is getting a lot of traction in the Factory Automation market, with more than 20 customers that are both prestigious and demanding. We successfully performed more than 150 feasibilities studies in less than a year. That was possible because it really only takes minutes to make a new application case from scratch. Our training classes are booked months in advance.