Hands Up Detection for Camera Review

Hands Up Detection brings a person with both hands raised to an operator for review.

Book a Demo

The Visible Condition

Describe What the Camera Can Show

The observable condition is a person with both hands raised at a bank lobby or retail entrance. A frame can bring the scene to an operator. It cannot establish intent, cause, or the right response.

A raised-hands prompt can bring an unusual posture to the control room. The operator still needs the wider view: people can raise their hands for ordinary reasons, and the same image set also depicts people on the floor. Keep those two visible conditions in separate prompts so staff know what matched.

An authorized operator can enter a plain-language prompt such as “Alert when a person has both hands raised in the monitored lobby.” StreamVLM™ evaluates sampled live frames from selected cameras against it. A match enters the IREX event flow.

Each detector has its own confidence threshold and alert cooldown. The team tests wording and settings on its cameras.

StreamVLM™ evaluates sampled frames, so a brief action can fall between samples. Test the actual camera view and response timing before relying on this prompt for an urgent event.

This prompt can be evaluated alongside the Weapon & Gunshot Detection trained module. Compare their frame-rate needs and operational fit during a site-specific pilot.

Explore StreamVLM™

Operational Fit

Plan the View and Review

Camera Placement

Use an entrance or lobby view that shows arms and nearby people. Check lighting, scale, and occlusion on site.

Operator Decision

The reviewer should check the surrounding scene for a welfare or security concern. The prompt does not decide this automatically.

Likely False Matches

Check for a greeting, stretching, or reaching for a shelf. Tune camera selection and alert settings during a pilot.

From Prompt to Action

How an Alert Becomes a Decision

  1. 01

    Describe and Select

    Use observable wording such as “Alert when a person has both hands raised in the monitored lobby.” on cameras that show the complete relevant scene.

  2. 02

    Review the Event

    Open the match and surrounding footage. Compare with ordinary scenes such as a greeting, stretching, or reaching for a shelf.

  3. 03

    Apply the Site Procedure

    Staff follow the site security procedure after reviewing the scene. Raised hands alone do not establish robbery or duress.

Pilot Questions

Validate on Your Cameras

Visibility
Can the view consistently show a person with both hands raised? Test representative lighting, weather, and occlusion.
False Alerts
Sample ordinary scenes, especially a greeting, stretching, or reaching for a shelf, then tune prompt, threshold, and cooldown.
Human Review
Agree who checks a match and how to check the surrounding scene for a welfare or security concern. No universal accuracy figure is published for this prompt.

FAQ

What does hands up detection report?

It reports a match to a described visible condition: a person with both hands raised. An operator reviews footage before deciding what happened.

Can a match establish intent or liability?

No. A sampled frame cannot establish intent, authorization, responsibility, or cause. Staff check context and follow local procedure.

Will raised hands alone identify a robbery?

No. The posture does not identify a weapon, threat, or criminal act. Review the broader scene and follow the site security procedure.

How should we evaluate it?

Test representative cameras and ordinary scenes, including a greeting, stretching, or reaching for a shelf. Agree on useful alerts and missed events. IREX does not publish a universal accuracy figure for this prompt.

Test Hands Up Detection on Your Cameras

Choose representative views of a bank lobby or retail entrance. We can scope the prompt, alert settings, and review procedure for a pilot.

Book a Demo