Developing Embedded Vision

The field of embedded vision is growing rapidly with many cool vision-enabled products coming. If you’re involved with a product using embedded vision, here are a few helpful resources to bookmark:


The Embedded Vision Alliance (EVA): On the website for the EVA ( you’ll find lots of useful information to help you make the best embedded vision product possible. They have info on the latest news and trends in the industry, presentations from the recent Embedded Vision Summit and even free online training at the Embedded Vision Academy section. TI is proud to be a founding member of the EVA and you can find a lot of information on TI’s demos and libraries on the EVA site.


TI’s Vision Library: If you’re one of the many people using one of TI’s DSPs for embedded vision, then make sure you are familiar with our vision library (VLIB), which is available for free download here. Our VLIB has optimized building blocks for common vision functions like background modeling and subtraction, feature extraction, tracking, recognition and low-level pixel processing. Due to the need for high-performance, most customers in embedded processing prefer to use optimized routines like our VLIB routines rather than OpenCV calls.


Dream DSP: While not all the information here is geared specifically towards embedded vision, the DreamDSP page is a great source for information. You’ll find the latest news and information on TI’s DSPs, which are great processors for embedded vision applications. From here you can download our paper on embedded analytics, see TI Designs that can get you started, and watch videos showing how DSPs are used for analytics and TI’s role in the DSP space.


Speaking of videos, we have a new DSP Breaktime video out focused entirely on embedded vision. In this episode, Arnon and I discuss why DSPs are good for embedded vision, the Embedded Vision Alliance and OpenCV on TI’s processors. We even relate our movie topic to embedded vision.

Be sure to watch episode 4 and tell us about your cool embedded vision application.