Home » Argu Products » Facial Recognition

Facial Recognition

Facial recognition security deployed on premise, so biometric matching happens on your own infrastructure and identity data never has to leave the site.

The differentiator is where the data lives and how the capability is used. Facial recognition is optional, switched on only where a site needs it, and built into the same agent workflow as every other capability rather than sold as a standalone biometric silo. Deployment is on premise, so biometric matching happens on infrastructure the customer controls, configured to local privacy and access law. Confirming identity at a controlled entrance is one piece of a broader homeland security surveillance posture.

Key facts:

Both 1:1 verification and 1:many identification run from the same on-premise deployment, consistent with the intent of GDPR Article 4's definition of biometric data as a distinct, sensitive category.
On-premise processing keeps identity data on infrastructure the customer controls, relevant at sites already restricted under NDAA Section 889 or using FIPS 201/PIV identity credentials.
The capability pairs with existing gate hardware over RTSP, ONVIF Profile S/T, IP camera, and NVR connections, without a dedicated biometric reader.
Matching runs on standard camera feeds at 1080p–4K resolution and 15–30 fps, the same range used across Argu's other vision-agent capabilities.
In US jurisdictions, biometric identifiers fall under statutes such as Illinois BIPA (740 ILCS 14) and CCPA, both of which set consent and disclosure requirements independent of Argu's own deployment.
A GDPR Article 35 data protection impact assessment is the standard compliance step for biometric processing in EU jurisdictions, and on-premise deployment is configured to support that review rather than bypass it.

Facial recognition folds into the same homeland security surveillance posture referenced above, and casino operators run it as one layer inside a wider casino surveillance program. At the gate it pairs with tailgating detection and access control monitoring, and sites already running video redaction apply the same on-premise privacy posture to both capabilities.

Axis	Conventional cloud biometric silo	Argu on-premise vision-agent approach
What raises the alert	A 1:1 or 1:many match returned by an external vendor's API	A match confirmed on infrastructure the customer controls, inside the same agent workflow as every other capability
What a human must watch	A separate biometric dashboard, apart from the main surveillance feed	One alert stream alongside every other Argu capability at the site
Integration path	New biometric-specific hardware or a cloud API contract	Existing RTSP, ONVIF Profile S/T, IP camera, and NVR connections already in place
Applicable standards context	Biometric data leaves the site, subject to a third-party processor's own terms	Consistent with the intent of GDPR Article 35's biometric impact-assessment requirement and Illinois BIPA (740 ILCS 14) consent provisions, since data does not leave the facility
Camera / feed required	A dedicated biometric-grade camera at the reader point	Standard 1080p–4K feeds at 15–30 fps from cameras already mounted at the gate
What gets retained	Match logs on a third-party server	On-premise identity records retained per the facility's own privacy and access-control policy

What is Facial Recognition

Facial recognition security here is optional, an enhancement inside Argu's agentic workflow rather than a separate system a team has to manage on its own. It supports 1:1 verification, confirming that an employee at a gate is who they claim to be, and 1:many identification, locating a specific person across live or archived footage. Every deployment runs on premise, and configuration must follow the privacy and access control law that applies to that site.

Key Takeaways

These are the core principles that define how Argu approaches every deployment. Our solutions are built to work on your existing infrastructure, require no engineering overhead, and deliver measurable results from day one. Every feature is designed with one goal: giving your team the intelligence to act faster, with more confidence, and without adding complexity.

See it in Action

Identity confirmed at the gate, on premise

Specific person located across footage

Biometric data never leaves the facility

How It Works

Decide where facial recognition applies

A security lead identifies which gates, doors, or zones need 1:1 verification or 1:many identification, since the capability is optional by design. That scoping decision is what keeps the capability optional rather than a blanket default across every camera on site.

Reference identities are set on premise

Authorized identities are configured on infrastructure the customer controls, in line with the privacy and access control law that applies to the site. Reference data is stored on the customer's own infrastructure, not synced to an external vendor's database.

The agent matches faces at the gate or across footage

Depending on configuration, the agent confirms a known identity at an entry point or searches for a specific person across live and archived footage. Matching runs on standard camera feeds at 1080p-4K and 15-30fps, without a dedicated biometric reader.

Access or alert decisions stay with your team

A match produces an alert or an access decision for your team to act on. The agent does not grant entry or make the final call on its own. That human-in-the-loop step is also the point where a facility documents its own access-control policy for audit purposes.

Product Advantages

Frequently Asked Questions

Does facial recognition data ever leave our facility?

No, as configured. Deployment is on premise, so biometric matching runs on infrastructure your team controls rather than an outside server. That on-premise design is also the deployment condition needed to support a GDPR Article 35 impact assessment for biometric processing, since data never reaches a third-party processor.

What is the difference between 1:1 and 1:many identification?

1:1 verification confirms that a specific person, such as an employee at a gate, matches the identity on file. 1:many identification searches for that person across live or archived footage instead of checking one identity at a time.

Is facial recognition required to use the platform?

No. It is an optional biometric enhancement, switched on only where a site needs it, and it runs inside the same agent workflow as every other Argu capability.

Does this work with our existing access control and cameras?

It integrates with existing gate hardware and camera infrastructure over RTSP, IP camera, and NVR connections, without requiring a separate biometric system on top. Feeds run at standard 1080p-4K resolution and 15-30 fps over RTSP or ONVIF Profile S/T, the same range used across every other Argu capability.

How do you handle privacy and legal requirements across regions?

Every deployment is configured to the privacy and access control law that applies to that specific site, and the capability is only activated within that framework. In the US that can include statutes such as Illinois BIPA (740 ILCS 14) and CCPA; in the EU it can include the GDPR Article 35 impact-assessment requirement for biometric processing.

What resolution or frame rate does facial recognition need from our cameras?

The capability works with standard 1080p-4K feeds at 15-30 fps, the same range Argu's other vision-agent capabilities use, so a facility does not need a dedicated biometric-grade camera at the gate.

Does facial recognition replace our badge or key-card system?

No. It runs alongside existing credential hardware, such as badge readers using Wiegand 26-bit or OSDP v2 protocols, adding an identity check rather than replacing the credential itself.