Paravision brings edge-to-cloud facial recognition to AI digital ID disruption party
Paravision’s ‘Converge’ 2021 announcement includes five separate computer vision and biometrics product releases and updates, covering a wide range of possible applications and industry verticals.
Paravision Chief Product Officer Joey Pritikin and President and Chief Operating Officer Benji Hutchison joined Biometric Update to discuss what the ‘Converge’ mega-launch means for the industry.
After taking years building up to it, the release is “not a ‘move fast and break things’ moment,” Pritikin cautions.
Pritikin says that launching the tools and facial recognition capabilities at once can help customers understand how they can use Paravision’s portfolio. The launch will “give them an opportunity to synthesize it, so that they can think through how all of these different toolsets and capabilities and technologies and products can come together to help power what they’re building.”
This is necessary, they say, because Converge is the culmination of several years of work on technology, customer and market development, and introduces capabilities that are new to the company, if not the industry.
Some of the major players thinking about cutting edge approaches Paravision is taking, Hutchison says, but perhaps because they are not as nimble, they have yet to put them into practice. Pritikin refers to the advantage that Paravision has over many of the larger incumbents in the biometrics space as a “dynamic” quality, while Hutchison refers directly to the company’s smaller size.
Ethical facial recognition
“Global perspectives on AI in general and face recognition specifically are helping us to define and refine a vision for ethical development and use,” Pritikin says. “So features, benefits, price; that’s no longer sufficient to explain a product offering.”
Asked about ethical facial recognition development, he refers to Paravison’s focus on inclusivity, and on being aware of biases introduced through data, tech or culture.
Pritikin praises NIST FRVT as a unique transparency tool. “It is perhaps singular in the world in terms of providing comprehensive, transparent insights into how AI-based products work,” he says. “I mean think of it; 1:N, 1:1, demographics, facemasks, varying database sizes, performance over time, a huge variety of image types.”
Sales ethics are related as well, Pritikin states. Paravision has firm guardrails established with Chief AI Ethics Advisor Elizabeth Adams.
The company also holds weekly use case reviews to screen all new business, and turns down opportunities. Paravision will not sell to the dozens of nations considered countries of particular concern by the U.S.
“It’s hard to turn away business sometimes,” Hutchison admits.
“If we can’t find the necessary legal precedence, the frameworks, the privacy law . . . we’ll walk away,” he elaborates a moment later.
He sees industry best practices being put into place, and says Paravision is trying to provide leadership in the area.
A convergence of trends
Just prior to COVID, Hutchison observes, edge was still a server sitting near the end device, not driven by AI, in the minds of most, even among the few who were talking about the concept.
“Some of the topics that we bring to the table are new for a lot of people,” he acknowledges.
“That’s what you’re seeing out of the Converge announcement,” Pritikin explains. “Edge AI needs to take into account not only the software toolsets – and we’ve put a lot work into making those easy to use – but also things like hardware reference designs.”
Hutchison describes the Scaled Vector Search product as “a modern search engine with facial recognition one-to-many.”
After joking about pronouncements about ‘the year of biometrics’ a decade ago earlier in the conversation, Hutchison expresses confidence
“When you see the breadth of the Converge announcement, it’s really about the ability to enable them with world class face recognition and associated computer vision throughout their solution or service that their developing,” Pritikin sums up. Converge also refers to the technology bringing together different domains, such as physical and logical access control.
Hutchison sees that step as part of a broader trend playing out across the entire digital identity industry.
“For the longest time, biometrics was dominated by fairly niche players and systems integrators and a handful of very large companies, and I think that you’re seeing that disrupted now,” he says. “I think that you’ve seen over $2 billion be injected into the identity space over the past year, and we’re a part of that.”
“It’s been a long time coming.”