Today, a great deal of effort is wasted in optimizing and re-optimizing computer vision and machine learning application software as algorithms change and as developers target different processors. Developers need a way to deploy their applications on any processor, taking advantage of processor features that boost performance without having to worry about the details of mapping each algorithm to each processor. OpenVX is a mature, open and royalty-free standard for cross-platform acceleration. It enables computer vision and machine learning applications to be written once, and then run on a variety of target processors, taking advantage of each processor’s unique capabilities. 

In this talk, we’ll explore the key features of OpenVX 1.3.1 and show how they are being leveraged by developers. We will illustrate the performance, portability and memory footprint advantages of OpenVX via open-source code samples. We will also share new OpenVX extensions and usability enhancements.