Zoom and Enhance: Imagination Technologies at the Embedded Vision Summit

IMG_2905-1024x768

This blog post was originally published at Imagination Technologies' website. It is reprinted here with the permission of Imagination Technologies.

In our report on last year’s Embedded Vision Summit, we observed that machine learning was at the top of the classic Gartner hype cycle. Imagination was of course once again in attendance at the show last month and we’d say it was still riding high.

Kristof Beets, Senior Director of Product Management & Technical Marketing at Imagination reported back that there was a real buzzy vibe from the show and that it was busier and more exciting than ever.

“The show was definitely bigger this time around and what we saw had a very technical, raw feel to it. There were many things on show from a wide variety of start-ups, including several from China, but it was interesting that in the main they were technical demos based on FPGA systems and test boards rather than slick demos based on finished products.”


There was lots of activity and interest at this year’s EVS show.

The increased size of the show was certainly backed up by the numbers from the organisers who reported that there were more than 1,000 attendees, more than 90 speakers across six presentation tracks and more than 50 exhibitors demonstrating more than 100 vision technologies and products.

It’s clearly a competitive marketplace but this helped give the show an exciting, buzzy feel that keenly demonstrated that this is a highly competitive area.


Russell James, Vice President of Vision & AI at Imagination demos object recognition on PowerVR at the Embedded Vision Summit 2018.

One new area that was seeing a lot of work Beets noted was around PoseNet, a neural network framework that is enabling smart new efficient ways of determining human figures in images and videos – great for gesture recognition, for example.

There’s also work going on around using neural networks for creating ‘super video resolution’ for image enhancement and even upscaling video to 8K. Who knows? Maybe using these techniques reality will finally catch up with the ‘zoom and enhance’ TV and movie trope.

 

Francisco Socal, Product Manager, Vision and AI agreed that there were some very good discussions with a good combination of technology, research and business insight. “There were many discussions about data and impact from GDPR in Europe even if many AI players are US-based.

“What I found interesting is that the show managed to see beyond the hype of AI, to capture what really matters and is actually being deployed in real-world problems and commercial applications,” said Socal.


Some might say they like to have their ducks in a row but as a graphics company, we’re going for the 3D stacking approach.

Socal also said the show benefited from some inspirational speakers in the guise of computer vision pioneer Takeo Kanade, inventor of the Lucas-Kanade optical flow algorithm and currently U.A. and Helen Whitaker Professor at Carnegie Mellon University. Some choice quotes were that those in the AI field should, “think like an amateur, but do as an expert,” meaning that we should always use good science to ensure that our AI solutions help solve real problems and that we don’t get carried away with this new hype-filled technology for the sake of it. “Newness itself is not all, usefulness is,” Kanade said.

The other keynote speaker was Dean Kamen, the founder of DEKA Research and Development. He pointed out that there have been a lot of serious concerns raised recently around privacy in the IoT and the many smart devices now proliferating in homes but that AI could play a major role in helping deal with this problem. “We can protect privacy even as smart cameras proliferate by baking in privacy, and embedded AI can do exactly that,” he commented.

Imagination contributed to the discussions with a talk from our senior research manager for PowerVR Vision & AI Paul Brasnett, who spoke at the event on the subject of: “Implementing and improving rational computer vision algorithms using DNN techniques.” Look out for coverage of his talk here on the blog in the coming weeks.

With our range of PowerVR GPUs and our recently announced Series2NX NNA processors for both the premium and mid-range markets, Imagination, of course, is ideally placed to make good on the promise of the latest trends and techniques in AI. As Beets observes: “many companies have good technology but to implement it in a product you need technology from a company that’s well-funded, experienced and competitive to move it forward”.

Sounds a lot like Imagination I would say.

Benny Har-Even
Technical Communications Specialist, Imagination Technologies

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