Biometric Facial Recognition

ROC facial recognition software analyzes an image or video, detects the face, and generates a compact biometric template - a numerical encoding that cannot be reverse-engineered into a photo. That template powers two core operations:

• 1:1 verification - are these two images the same person?
• 1:N identification - does this person appear in a database of millions?

Vendors differ significantly in accuracy, speed, and fairness - which is why NIST FRVT exists as the independent government benchmark. ROC is the only American-made provider ranked in the global top tier for both accuracy and efficiency across all NIST test categories.

Source: NIST FRVT — More: roc.ai/testing-and-standards

ROC is the only American-made facial recognition software ranked #1 globally in combined speed and accuracy by NIST FRVT. Three enterprise-grade differentiators set ROC apart from every competitor:

• Hardware-agnostic - no proprietary lock-in; runs on any server, edge device, or mobile hardware
• 38MB base library supporting galleries of 500M+ identities, with templates 160x smaller than nearest competitors
• Single SDK, 14 biometric modalities - face, fingerprint, iris, liveness, ALPR, gun detection, and more in one integration

Deployed by the U.S. DoD, FBI, and U.S. Marshals. Powers ~50% of the global ID-proofing market.

More: roc.ai/sdk

A biometric face template is a compact numerical encoding of facial features - not a photograph. It cannot be reverse-engineered into an image, which protects against identity spoofing.

ROC stores templates, not raw images. All data is encrypted in transit and at rest, access is policy-bound and auditable, and retention periods are configurable per organizational requirements. The face blurring feature protects the identity of anyone who is not a person of interest in video analytics deployments. ROC complies with applicable biometric privacy laws across every jurisdiction where it operates.

More: roc.ai/code-of-ethics

All three deployment models are fully supported with identical algorithm performance:

• Cloud - elastic scaling for city-scale or national identity programs
• On-premise - full data sovereignty for agencies with strict residency requirements
• Edge - the 38MB base library runs on mobile hardware, IoT devices, and patrol laptops without cloud dependency

Air-gapped deployment (no internet connection required) is fully supported for classified government and military environments. One SDK covers your entire infrastructure - no separate codebase per deployment type.

ROC serves five core verticals through purpose-built configurations of the same NIST-ranked algorithms:

• FinTech and Digital Identity - ~50% of the global ID-proofing market, including eKYC, account opening, and anti-fraud platforms
• Law Enforcement - 12+ U.S. agencies, FBI, and U.S. Marshals for RTCC, forensics, and ABIS
• National Security and Military - DoD deployments for perimeter protection and field identity operations
• Government and Border Control - national ID programs in 50+ countries
• Healthcare - patient identification and secure facility access

More: roc.ai/digital-identity — roc.ai/national-security

NIST FRVT independently confirms that ROC algorithms exhibit near-zero differences in error rates across racial and gender cohorts - this is a benchmark result, not a marketing claim.

ROC's approach to fairness includes:

• Training datasets that are ethically sourced and precisely annotated across diverse global populations
• Continuous bias monitoring built into every model release cycle
• An industry-leading Code of Ethics governing how ROC technology may and may not be used

Source: NIST FRVT demographic report — More: roc.ai/code-of-ethics

ROC defends against presentation attacks through single-frame passive liveness detection - a US-patented technology using micro-texture analysis that confirms whether a face is real in under 1 second, with no user action required.

Protection covers all major attack categories:

• Printed and digital photo attacks
• Video replay attacks
• 3D silicone mask attacks
• AI-generated deepfake face presentations

ROC Liveness is iBeta Level-2 certified (ISO 30107-3) - the highest commercial standard for Presentation Attack Detection. Works with any standard RGB camera - no specialized hardware required.

More: roc.ai/liveness — Source: ISO 30107-3

ROC provides a unified, cross-platform SDK with a standardized API interface that eliminates the need for modality-specific re-architecting. Feature parity is maintained across 7 languages—C++, Java, Python, C#, Go, Node.js, and Rust—with native support for Windows, macOS, Linux, iOS, and Android.

Integration Options:

• REST API - Engineered for modern web services and microservice-heavy architectures.
• Native SDK - Optimized for C++-level performance in high-throughput embedded and edge applications.
• CLI - Ready-made for automated batch workflows and shell-based processing.
• Web API - Delivers robust server-side deployment with granular hardware acceleration control.

ROC is built on open standards to ensure full interoperability and eliminate vendor lock-in. Most enterprise-grade integrations move from evaluation to production within days. Start building immediately with a free 30-day evaluation license, which includes complete documentation and direct access to our senior engineering team.

More: roc.ai/sdk

ROC's largest live production deployment is a U.S. federal government program with a 500M-template gallery operating at mission-critical speed. The architecture scales without reengineering:

• 38MB base library - compact enough for edge devices, fast enough for cloud clusters
• Smallest templates in the industry - reducing storage costs at national scale
• Horizontal cloud scaling natively supported
• Sub-second search latency maintained across 100M+ gallery sizes per NIST benchmarks

From a startup's 10,000 users to a national ID program - the same SDK handles both.

The ROC SDK is available for Windows, Mac OS, iOS, Android, and Linux. The ROC SDK supports both x86-64 and ARM processors. For older processors we provide a legacy build for CPUs that do not have the FMA instruction set.

Yes, ROC is very accurate when used on children. However, there are important nuances to this answer.

1. Rapid physiological development – Early adolescence is a time in physiological development when the face and cranium experience major changes.
2. Availability of data – Face Recognition algorithms exhibit high performance on population segments on which they have been trained. Training data for minors is not as readily available as data for adults.
3. The time span between enrollment and subsequent matching is especially important to consider when Face Recognition systems are used with younger populations. Overall, we estimate that Face Recognition works well on people over the age of 11.

Thanks for asking! We are extremely proud of our Face Recognition performance, which is at the core of who we are as an AI/ML company. The ROC Face Recognition algorithm consistently ranks in the top tier of accuracy, template size, and enrollment speed of algorithms in the world according to the NIST FRVT. We also conduct regular internal benchmarking, which affirms our ability to turn around new and improved algorithms fast.

We recommend consulting the NIST FRVT 1:1 and 1:N reports, which presents a full-picture of each Face Recognition algorithm vendor’s performance, as well as how they stack up against each other.

ROC believes that bias in facial recognition is not a topic that should be ignored. We have made it a priority to limit bias within our algorithm as much as possible, and it shows. Our technology continues to exhibit minimal differences in error rate between racial and gender cohorts. This blog post has more information (section on “Algorithmic Bias”). We are constantly working to improve the accuracy of our algorithms, including reducing gender and racial bias.

ROC believes that preserving privacy should be a primary consideration of facial recognition vendors. We have led the industry in adopting a comprehensive code of ethics and advocating for ethical use of AI across the board. ROC strictly complies with any and all laws and regulations relating to privacy, data security, data protection, and processing, transfer, disclosure, sharing, storing, security and use of personal information, as applicable in any jurisdiction where it is used. In addition, ROC has developed features like “face blurring” which protect the identity of all except persons of interest.