AnyVision on Cracking AI Before It Was Fashionable
An Israeli visionary entrepreneur, a world renowned professor in computer vision, and an ex-banker walk into a room. No, it’s not the start of joke. These are just three of the many brilliant minds behind AnyVision, an AI powerhouse that’s quickly drawing the attention of many. And one of the many is Qualcomm Ventures.
Today, we’re proud to announce the creation of Qualcomm Ventures’ AI Fund, a commitment to invest up to $100 million in companies transforming AI with an emphasis on the edge. This fund will build on our demonstrated track record of investing in top global AI startups.
As part of this fund, we made an investment in the Series A funding round for AnyVision, a world leading face, body and object recognition startup. We believe AnyVision’s use of on-device AI is quite innovative as it uses technology to run video analytics on the device, therefore mitigating privacy concerns.
In celebration of the new AI Fund and our investment in AnyVision, we sat down with Co-Founder and CEO Eylon Etshtein, Co-Founder and CTO Professor Neil Robertson and Chief Commercial Officer Max Constant (the aforementioned “three”) to chat about what makes the company so unique (hint: those brilliant minds).
How did AnyVision begin?
Professor Neil Robertson: Back in 2011, I was working with one of my PhD students, Rick Mukherjee (now the company’s Head of Research), on deep learning — before it was fashionable. Eylon had invited him to collaborate on a face recognition project, and we were coming up with some interesting stuff. Things looked like they were going in an interesting and potentially very successful and impactful direction, so we got onboard with Eylon. We formed a very strong working relationship focused around a particular problem we’re very passionate about solving. And, as they say, the rest is history. But I’d be interested to hear what Eylon has to say.
Eylon Etshtein: When we started the company, I had been searching for the top minds in AI. Professor and I share the same passion, the same philosophy about what we can do to make this world a better place. We may have different backgrounds, but the beauty is that we have a collective vision for doing what we do.
Would you tell us about your products, the technologies they’re using and their applications?
Eylon: Better Tomorrow — we named it that because we actually want to create a better tomorrow — is an advanced tactical surveillance system that’s powered by a deep neural network. And there are all kinds of use cases: airports, cities, banks, casinos. Security cameras are everywhere, and unfortunately, the world can be a dangerous place. Our solution provides real-time face recognition so basically, after 30 minutes of installation time and training, you can start protecting the people that you care about. It’s magic.
SesaMe, our authentication and deep learning platform, makes identifications using a very unique algorithm that this brilliant mind, Professor, discovered. Essentially, we train the reflection on the screen on you to create a 3D representation of your face. Then we can authenticate and say you are who you say you are in 0.2 seconds. One of the pioneer clients of this solution is Bank Hapoalim, a leading bank based in Israel. Today, its customers can open an account without the need to physically go to the branch.
There’s another product we’re very proud of called Insights. Insights is not security oriented, it’s more analytics based. It’s about understanding the customer experience, the shopper experience. Take, for example, big retail stores. Analytics from their mobile app and website give them all they need to know about their customers. But when it comes to their physical branch, they know nothing about the customer or shopper until a purchase is made. The beauty of our technology is that, from sensors or cameras we gain insight into shopping patterns, which can determine things such as which brands are popular, etc.
How is your technology unique?
Neil: I think there are two things to understand that separate AnyVision from other AI companies: our algorithm is designed for the real world and we take a holistic design of hardware and software.
We actually went after really hard underpinning problems in the AI space. We recognized early on that, if we could design an AI algorithm that would take away the burden of having to replace your sensors or worry about lighting conditions or which angle you were pointing your camera, there would be enormous market interest — and an enormous impact in terms of actually solving a lot of core problems.
We use the cameras that people already have to do facial, body and object recognition. That’s what Better Tomorrow does. We take a very clever approach to the way you structure the data. We have a very strong technical team that can work out how to do the machine learning. We also remove the redundancy and a lot of competition. This allows us to put these solutions out there into the world — onto phones, onto very lightweight systems that can run 30, 40 cameras off of a single GPU.
Is that where Qualcomm comes in?
Neil: Yes, because it was clear that Qualcomm was pushing things further and further to the edge. We were one of the few AI companies that had the ability to actually design algorithms for very low-power, small-scale computing chips. It’s exciting because the more integrated we are, the closer we can get to the actual hardware design. The better, more efficient our algorithms can be, the more seamlessly they can work with Qualcomm chips.
Eylon: The beautiful thing about the synergy with Qualcomm and their investment in Anyvision is that it has basically brought us to the platform world. We also discovered that we need to collaborate with a company that’s coming from the silicon industry or someone who’s coming from computing. We’re really happy that Qualcomm recognizes the technology breakthrough that we’ve made with the Qualcomm chip, that we expect it to change the world and be a critical part of us making our technology accessible.
What sets AnyVision apart from the competition?
Eylon: Besides the amazing technology and the brilliant core team that’s behind the technology, I think it’s how we go to market. Lots of companies create one product and they sit in one vertical, and that’s it. They’re a product company. We are not. We are a technology company. We want to be the leader in AI.
What’s next for the company?
Eylon: By the end of 2019, we want to be in a place where we’ve finished phase one in the structuring of the company. Phase two is going to be, I think, more futuristic in terms of the targeted platform that we’re going to be able to create. From the technology standpoint, we’re basically on a journey to index and analyze everything the sensor sees, and once we do that, we’ll be able to solve even more fundamental problems, let’s say, on the pixel level.
Neil: I think from a technology point of view, we went hard after one of the most studied problems in computer science: face recognition. The insights that we gained about machine learning and cracking that problem make some other problems very easy to go after. That’s what Eylon’s talking about — that you can learn all kinds of patterns from video, which means that you can start to build up very clear pictures of what’s in the world and relationships between objects and start to put together very powerful analytics. That’s where we’re going.
What’s your prediction for AI next year and then a little further out, say, five years from now?
Neil: I think from a technology point of view, there’s an interesting thing happening. The most important streams of the computing spectrum are under pressure. What I mean by that is that people want functioning AI on their devices. They want it in their car. They want it in their home. Small-scale computing at the edge is being driven more and more as a potential solution to deliver some of these algorithms.
I think we’ll see more and more companies attempting to put high-powered AI onto edge devices as a business strategy, but I think many of them will fail because what they’re trying to do is reverse engineer solutions that weren’t designed for that purpose. I think this year as well we’ll see the emergence of AI as a service. So, at the other end of the computing spectrum, we have very large computing systems on the cloud. That will be the future because of the vast volumes of data that are being generated, and also because we need to distribute the services across internet scale-level applications. So, I suspect that the desktop or small lab scale of AI will start to disappear this year.
Max Constant: Agree. I think it’s a very misunderstood shift in the market. From a timing and technological gap perspective, I think we are leaps and bounds ahead. The hard work from Professor and his team and the investment by Qualcomm is testament to what we’re able to achieve and how quickly this industry is coming.
What is it that excites you about having Qualcomm Ventures as an investor and what are you hoping to gain from the relationship?
Eylon: First of all, to explain AI is not easy. We encountered amazing people that really understand what they’re saying from the tech perspective and what we’re doing. For us, as entrepreneurs and software guys, it’s really fun to speak the same language.
We feel that Qualcomm Ventures has a very cutting-edge approach in terms of where they want to be and the companies they will put inside the portfolio. We are really glad and grateful to be the first investment in the AI Fund. And it was very exciting to see all the portfolio services that Qualcomm offered us in order to become a better company.
Max: To build on what Eylon said, there’s great support and initiatives that we’ve learned about. Then there is the access to the broader Qualcomm ecosystem. Since we view ourselves as a core technology provider to the industry, it’s really important for us to have industrial validation, industrial partners, like Qualcomm, that are really at the bleeding edge of where the industry is moving in two, three, five, 10 years. We use the term “bottled lightning” internally. Meaning: It’s a rare opportunity when you can actually go and create an industry with the biggest industrial partners in the world. I think having Qualcomm Ventures as a very early believer in what we’re doing is really important moving forward. We’re proud to have that as part of our story. And it’s just the beginning.
Eylon: AnyVision is going to be the leader in this industry using Qualcomm edge computing.
The opinions expressed in this piece are solely those of the interview subjects and do not necessarily reflect the views of Qualcomm Ventures.
