Video: Feature tracking and optical flow at the Embedded Vision Summit


Last October we demonstrated our computer vision processor at the Embedded Vision Summit in Boston. Among others, we showed our Harris-based feature detection running on HD resolution camera input, as well as an optical flow implementation, using the Lucas Kanade algorithm. On such algorithms we often get a 100x speedup compared to a host CPU, and a factor of 1000x better power efficiency. Watch Mark Kulaczewski demonstrate our real-time HD vision system:

Do you want to see all our demonstrations or have any questions? Please contact us. We’re often traveling around the globe so can set up a meeting or schedule a call.

16/01/2014 / Marco Jacobs