Skip to content
Camzify
Use case · Schools and civic buildings

Weapons detection for schools and public buildings

Weapons detection for schools and public buildings is the visual detection of a firearm or an edged weapon that is visibly held in a camera frame, raised as a critical alert with a clip to a person who verifies it. It runs on the entrance, corridor, lobby and parking cameras a school, a campus, a courthouse or a city hall already has. It is not a screening system, and it does not decide anything: a person looks at the clip and starts the building's own procedure.

Book a demo Weapons detection
  • Visible firearms and blades, raised as critical
  • On the cameras already at the doors
  • Not a scanner, not facial recognition
The console live view of a lobby by a revolving door, one person outlined with a Gun Detected label while others walk past
The problem

The plan exists; the question is when it starts

Nearly every US public school has a written active shooter plan, and most civic buildings have a version of one. The plan starts when someone knows. In practice that is a shout, a call to the front office or a fire alarm pulled in a corridor, and the minutes before that are the minutes the plan was written to use.

The entrance camera usually sees the weapon before anyone in the building does. It records the frame for the investigation, which is the wrong moment to be useful.

Screening at the door, where a building has it, covers the door. It does not cover the parking lot, the athletics field, the evening rental of the gym, the loading dock or the public counter that was never screened. Those are covered by cameras that nobody is watching.

How Camzify handles it

Raised when it is visible, verified by a person

Weapons detection runs on every enabled camera and evaluates each tracked person in the frame.

  • A visibly held or brandished firearm or edged weapon is the event.
  • The alert defaults to critical severity and carries the frame, a clip and a confidence score.
  • Repeated detections on the same tracked person are one event, not a stream of alerts.
  • The person receiving it acknowledges it, and the record shows who and when.

Detection runs at every hour. A notification window per camera limits when notifications go out, not when the detection runs, so the school day can notify the front office and the evening rental can notify the district operations center. The notification settings cover channels, severity and escalation per camera.

Aggression and fight detection is usually enabled alongside it, because one escalates into the other. Afterwards, AI suspect search finds the same person on other cameras by description, and cloud backup holds the footage within the retention set for that camera. The broader violence and weapons detection page covers both detections across other settings.

Detection

Weapons detection

A visibly held firearm or blade in frame, raised as critical with a clip and a confidence score. Not concealed weapons.

Detection

Aggression and fight detection

Sustained aggressive movement between tracked people in a corridor or a lobby. A crowded hallway at the bell does not match.

Detection

Multi-object tracking

Each person followed as one track, so the weapon model attaches to a person and repeated detections stay one event.

Detection

AI suspect search

Find the same person on other cameras afterwards, by clothing and color. Not facial recognition.

The patrol round

What an entrance round checks

Weapons detection is continuous and does not wait for a round. The round records the conditions the plan depends on: at each stop, the exterior doors were secured, the vestibule was staffed, the restricted corridor was empty, the camera could see. Repeated through the day, it is a record that the building was in the state the plan assumes, which is the record a district or a facilities department is asked for afterwards.

A failed item messages the person designated for that camera and stays Pending in the report until it is marked Fixed. On an automated round, the AI also raises a critical notification for a risk it sees at a stop that the checklist did not ask about. Unauthorized access detection covers the propped door and the tailgated entry that the plan also assumes are not happening.

A check found Not Compliant captures a snapshot and messages the guard designated for that camera; on a manual round the operator chooses to send it, on an automated round it goes on its own. The item stays Pending until it is marked Fixed, which captures the after frame, and the report shows both.

The US context, in published figures

Figures from the FBI and from the Department of Justice and Department of Education joint report on school crime, quoted as the source states them. They describe the country, not any one building, and none of them is a Camzify figure.

24
In 2024, the FBI designated 24 shootings as active shooter incidents, a decrease of 50% from 2023 (48 incidents). They occurred in 19 states across five location categories, including education and government.
Source: FBI, 2024 Active Shooter Incidents report (June 2025)
223
From 2020 to 2024, the FBI designated 223 active shooter incidents, in 43 states and the District of Columbia, a 70% increase from the previous five-year period (2015 to 2019).
Source: FBI, 2024 Active Shooter Incidents report (June 2025)
50
From 2000 through 2022, there were 50 active shooter incidents at elementary and secondary schools, with 328 casualties (131 killed and 197 wounded), and 18 at postsecondary institutions, with 157 casualties.
Source: NCES and BJS, Report on Indicators of School Crime and Safety: 2023
96 percent
From 2009-10 to 2021-22, the percentage of public schools with written plans for active shooter scenarios increased from 84 to 96 percent.
Source: NCES and BJS, Report on Indicators of School Crime and Safety: 2023

What it will not do

This is the page where the limits matter most, so they come first.

  • It will not detect a concealed weapon; it is a visual model, not a metal detector, a scanner or a bag check.
  • It will not prevent an incident; it shortens the time between a weapon being visible on a camera and a person knowing, and the response is that person's.
  • It will not call the police, start a lockdown or make any decision; it notifies the people configured for that camera and they decide.
  • It will not identify anyone, and it does not use facial recognition.
  • It will produce some alerts that a person, looking at the clip, will dismiss, and the workflow is built for that.
  • It will not see a corner without a camera, and a campus has many.

We do not publish detection rates or false-alert rates for the weapons model, and we will not put a number on a page about schools that we cannot verify for your cameras. The trust page explains why.

Industries where this applies

Education facilities Financial services Property management
FAQ

Frequently asked questions

No. Camzify weapons detection is a visual model that runs on the camera frame and flags a firearm or an edged weapon that is visibly held or brandished. It is not a metal detector, a millimeter-wave scanner or a bag check, and it does not see a weapon in a backpack, a holster or a coat. A building that screens at the door keeps screening; this covers the cameras beyond and around it.

Whoever is designated for that camera: the front office, a school resource officer, the security desk at a courthouse, a district operations center, or a monitoring company. The alert is critical by default and carries the frame, a clip and a confidence score. The person looks at the clip, decides in seconds, and starts the building's own procedure or dismisses it.

No. Camzify notifies the people configured for that camera on the channels configured for that severity, and a person decides what happens next. Whether that person calls 911, starts a lockdown or walks to the vestibule is the building's procedure, not ours, and we think the decision should stay with a person who has seen the clip.

Sometimes. The model is trained to reduce confusion with objects that look like weapons, and it does not eliminate it, which is why every alert includes a clip and a confidence score and why the workflow is verify, then escalate. A drama department rehearsing with a prop or a maintenance worker carrying a drill will produce alerts that a person dismisses.

No. Camzify does not use facial recognition and does not match anyone against a list. Detections attach to a tracked person in the frame, and AI suspect search can afterwards find the same person on other cameras by description, such as clothing and color. Nothing on the console names a person.

It will not detect a concealed weapon, it will not prevent an incident, and it will not replace screening, a threat assessment program or a response plan. It shortens the time between a weapon becoming visible on a camera and a person knowing about it, with a clip in front of them. The response, and its outcome, belong to the people on site.

Ready to patrol your site 24/7?

Book a 15-minute demo and see a live patrol run on your own cameras.

Book a DemoCalculate Your Savings

This site is being updated

We are rebuilding pages as you read this, so an image, a link or a section may look unfinished for a while. The product itself is unaffected. If something important is broken, tell us at the contact page and we will fix it.