Embedding Vision: Training Resources

EE_Times-logo

By Vin Ratford
Executive Director, Embedded Vision Alliance

This blog post was originally published at EE Times' SoC Design Line. It is reprinted here with the permission of EE Times.

This week, guest blogger Vin Ratford, executive director of the Embedded Vision Alliance, returns to discuss educational resources for meeting the challenge of programming heterogeneous processors. — Jeff

In my last blog, I talked about the challenges of learning to program the new generation of heterogeneous processing platforms and what the Embedded Vision Alliance and its members are doing to help developers harness the capabilities of these chips. Since then, I had an opportunity to attend the Vision Show in Boston, organized by the Automated Imaging Association (AIA). This event targets the machine vision industry and has a strong, well-attended educational program, including classes that lead to certification on machine vision applications.

The AIA has been developing standards that enable products to plug and play together. These activities make it a model for how industry ecosystems can work effectively. The AIA has been successful because it targets specific markets and includes suppliers, users, and system OEMs. I was impressed with the numbers of people attending AIA's classes and the educational content of the talks. I was a guest speaker; my session was well attended, and we had a lively discussion with lots of questions on embedded vision that centered on things like "How do I get started?" and "Where can I find out more about…"

As with all new technologies, embedded vision is moving beyond the innovators to the early adopters. Crossing the chasm requires education, improvements in enabling software, and a robust ecosystem. All these are interrelated and require a catalyst, which is exactly the role that the Embedded Vision Alliance has been fulfilling. We have found that educating the user community is key to industry success, and education is a key focus of the Embedded Vision Alliance.

The technical program for our Embedded Vision Summit, on Thursday, May 29, in Santa Clara, Calif., will revolve around two themes: visual recognition and autonomy. And we have expanded the summit education program with two workshops planned for Wednesday, May 28: an introductory workshop on design and implementation that will cover processors, sensors, algorithms, and development with OpenCV and OpenCL, and an advanced hands-on workshop that will focus on a specific hardware platform. For everyone from newbies to experts, these two days should help fill the knowledge gaps substantially and empower product creators to innovate using embedded vision technology.

Learning through startup

In addition to my role as executive director of the Alliance, I jumped in with both feet and started a company, Auviz Systems, which is developing accelerated middleware and IP for embedded vision applications on FPGAs. Our development team already had experience implementing vision algorithms on PCs, but reorienting our engineers to address heterogeneous processor platforms required learning OpenCL, FPGAs, and their tool flows while developing new ways to implement algorithms.

It has been a bumpy ride over the last six months, particularly with the tools, since the standards and APIs are new and evolving. In many ways, it reminds me of the early days of embedded software, when I was at Microtec Research, and the compilers and middleware for Motorola embedded processors BA (Before ARM) were not yet robust. Thus, I speak from personal experience when I say that companies will need to invest heavily in their development teams and anticipate lots of issues with heterogeneous platforms and tools for the next several years.

Regardless of these challenges, I fully expect to be "inside the tornado" (to quote Geoffrey Moore) with the Embedded Vision Alliance and Auviz in the next several years, helping to enable the next set of smarter, more capable systems and apps. Auviz will be at the Embedded Vision Summit Technology Showcase. I invite you to come see us there and continue the discussion.

Here you’ll find a wealth of practical technical insights and expert advice to help you bring AI and visual intelligence into your products without flying blind.

Contact

Address

Berkeley Design Technology, Inc.
PO Box #4446
Walnut Creek, CA 94596

Phone
Phone: +1 (925) 954-1411
Scroll to Top