Second annual product keynote shines light on expansion of Paravision’s product suite, including new image search capabilities and efforts to reinforce the future of trusted, authentic identity
SAN FRANCISCO — Oct 4, 2022 — Paravision, the leader in Trusted Vision AI, today announced new and updated products for its suite of leading Vision AI solutions at its annual product event, Converge. The event addresses leaders in sectors including identity verification, financial services, government, automotive and physical security as well as technology partners in Vision AI, chipset developers and camera manufacturers, showing the latest Paravision technology from cloud to mobile to the extreme network edge.
At Converge 2022, Paravision gave partners a look into its product strategy, which includes Face Recognition, Identity+, and broader Vision AI. For Face Recognition, the company announced a new enterprise-grade image search engine, Paravision Search. In Identity+, Paravision announced new products in advanced image attribute analysis and streaming video processing and privacy protection, and focused in particular on “authentic identity,” with new and improved liveness capabilities and a discussion of deepfake detection, currently under development.
Paravision has integrated its popular face-matching technology into a new enterprise-grade image search system, Paravision Search, which delivers massive scalability and support for very large databases and extreme request concurrency.
Paravision Search is easily configurable, and with its ability to maintain databases of hundreds of millions of records while enabling tailored sub-gallery searches, it is uniquely valuable for enterprise-scale use cases including identity verification, travel and borders, government identity programs and entertainment and events. Paravision Search also provides native elasticity that allows partners to adjust resources to meet real-time traffic demands.
We’ve taken industry-leading technology and packaged it in a way that better meets the scale, efficiency and reliability needs of the largest face recognition users from air travel to financial services. We’re able to integrate the latest technology and latest thinking from top-to-bottom, and of course API-driven and partner oriented just like our face recognition SDKs and Docker Containers.
Benji Hutchinson, Paravision President and COO
Last year Paravision shared its advanced Presentation Attack Detection for Ambarella CVflow, which uses near-infrared and other sensing to deliver real-time, edge-based PAD for a wide range of 2D and 3D spoof attacks. At Converge 2022, Paravision announced its first product for 2D RGB liveness and anti-spoofing, enabling partners to effectively check presentation attacks in standard 2D RGB images such as smartphone selfies. Deployable on mobile devices or server-side, Paravision’s 2D RGB Presentation Attack Detection can process single images in near real-time.
As a complement to its Presentation Attack Detection, Paravision recently announced a partnership with a trusted Five Eyes government agency to develop detection technology for digital face manipulations known as “deepfakes,” which pose serious risks to security, safety and privacy. With this new development, as discussed at Converge, Paravision will address a full range of face presentation attacks, whether digital or physical.
Deepfakes undermine trust in presented identity, and we’re already seeing this play out in entertainment, social media, videoconferencing and geopolitics. Our goal is to provide a trust layer for the physical and digital worlds to protect against malicious actors, fundamentally supported by an understanding of truth and reality in presented identity.
Joey Pritikin, Paravision Chief Product Officer
Paravision also unveiled its latest streaming video solution: Paravision Streaming 7. With Streaming 7, Paravision is extending its Vision AI for video beyond face detection to include person and vehicle detection, helping partners build solutions from perimeter detection to queue analytics to understand more about scene context.
Streaming 7 detection technology is complementary with a new capability under development: video redaction. Privacy concerns have become crucial when sharing data, especially within biometric applications. Redaction makes it possible to protect sensitive information at the source, and Paravision’s proven detection capabilities will enable partners to redact video for privacy-sensitive use cases, including faces, people, vehicles, license plates and more.
Finally, Paravision announced further enhancement to its Identity+ technology portfolio with an updated Attribute SDK, which supplements analysis of facial features with factors such as headwear, eyewear and masks.
We’re excited to address a range of needs our partners have requested, from ensuring authentic identity to protecting privacy to offering more insights from video. It’s truly rewarding to take the tools we’ve developed for face recognition, from technology to process to policy, and offer broader value to our partners in identity, safety and security.
Charlie Rice, Paravision CTO
This January, the National Institute of Standards and Technology ranked Paravision first in its rigorous 1:N Face Recognition Vendor Test (FRVT), making Paravision the first U.S. company to hold the top position in that benchmark since 2019. In April, the company was ranked first overall in face matching accuracy in the DHS Biometric Technology Rally, notably delivering 100% accuracy across all demographics. Paravision is now building upon these successes and learning from this domain to address wholly new areas in Vision AI.
We have so many opportunities to help advance science, healthcare, manufacturing, logistics, national security, agriculture and more. We believe most applications that require processing, characterization and understanding of visual information can benefit from Vision AI.
Doug Aley, Paravision CEO
Paravision’s Trusted Vision AI products are developed and sold under its AI Principles, which stipulate that its AI systems must be “ethically trained and conscientiously sold.” The company will carry these principles into any new markets or applications in Vision AI, setting the stage for considered, ethical solution development.
Watch the full video announcement here.