How to prevent school shootings

We can use camera-based AI to identify weapons -- even hidden weapons -- automatically and proactively, as described below. This is a modern way to reduce the risk of attacks in schools, offices, grocery stores, synagogues, mosques, and churches.

Today, we can go beyond monitoring social media accounts and placing metal detectors with human security guards in doorways. Sophisticated AI can be leveraged nationally and in my state, Colorado, where 47 have died. Colorado's incidents include Columbine HS (13 deaths), the Aurora movie theater (12 deaths, 70 injured), a Colorado Springs church (4 dead), Boulder's King Soopers grocery store (10 dead), and Highlands Ranch's STEM HS (1 dead) ~Denver Post . Incredibly sad.

As we know, this is a national problem. Here's a dynamic map , by the Center of Homeland Security, that shows 5 years of school shootings:

Some lives could have been saved if we proactively identified problems -- if we triggered early detection warnings. Noble efforts are being tried ... from monitoring Instagram and Facebook ... to installing metal detectors and by adding security guards. While nothing's perfect, AI can take proactive violence detection up a few notches.

AI plus a few cameras may be more effective than 1,000 Security Guards

Cameras with AI are extra safeguards. Enabling multiple cameras with AI puts eyes everywhere -- and the camera's eyes don't blink.

To me, using cameras and AI in schools would be like surrounding a school with 1,000 always vigilant security guards. Hundreds of people can be monitored at the same time as shown in this crowd and the AI can detect mask / no mask usage.

The cameras can detect "odd" masks and "off" behaviors -- 100s of them, all at once. The picture below shows an AI identifying a ski mask.

Seeing a ski mask in a commercial location or school could automatically lock doors and warn people inside the building of a potential threat. It could even call 911.

Below is an example of a real robbery where a person is detected wearing a ski mask and carrying a rifle.

Issues can often be detected before a person enters a building and before the violence begins.

Tapping into a school's existing video cameras, including ones that monitor the parking lots, can bring a whole new level of sophistication to reducing shootings.

The cameras can be easily linked to edge computing devices that process sophisticated AI algorithms on-premise and detect problems in 1/30th of a second. Their eyes continuously watch for threats like guns, knives, unusual behaviors, and masks that are different from the COVID masks.

One software that can run this Weapons Detection Application is Iterate.ai's Interplay, an AI-powered platform that is also low code and modular (pictured below).

It seems to me that a camera-based AI solution should become the first line of defense to protect our children from gun violence in public places. It can also be the first line of defense in grocery stores, churches, synagogues, mosques, theaters...

Proof points

We at Iterate.ai believe cameras can do the work. This belief is based on the experience we've gained over 18 months with roughly 8,000 cameras operated by a fuel station chain in Europe. For our fuel-store partner, Iterate built AI-visualization and payment tools that are compliant with GDPR and privacy laws. This program turns license plates into credit cards (pay-by-plate). It has rolled out successfully throughout Norway, Sweden, Denmark, Estonia, and Poland. Even at night, with sleet and snow, this AI hones in on license plates. It reads plates when cars are in motion, even driving at an angle. It then triggers a message to the driver's phone.

For license plate reading, we proved that monitoring dynamic shapes and regions optimizes the AI. It also binds the license plate to a gas pump in real-time, activates the pump automatically, and bills a credit card after pumping the gas. It works like magic.

Another proof point -- we are working with airports so they can comply with regulations. One has 5,000 camera installs. Those cameras create 10,000 simultaneous data streams. While an airport may require 30 to 40 edge servers to process these massive data streams, most schools would only require one.

New capabilities are emerging because cameras are getting smaller, better, and cheaper. Just look at what's happening in smartphone cameras. Advancements are being applied to cameras installed in businesses, too. They operate with WiFi, at night, with low power, in the rain, and so on. And they can be inexpensive.

For school security and weapons detection, the "brains" in those cameras (the AI) need to operate on small edge computing devices. Those small computers ingest the data streams and use AI to process the data in real-time. This results in automated intelligence and auto-alert systems.?

In our experience, it takes one edge device to run ~16 cameras. And that edge computing device should operate on-premise, meaning in a school's back-office. The edge devices process the AI algorithms. The AI algorithms are contained in Iterate.ai's modular low code platform, Interplay.

The fast path to protect students, grocers, worshippers, moviegoers ...

That said, Iterate.ai has trained its Weapon Detection App with a heterogeneous mix of AI techniques to create a highly calibrated engine that discovers guns, knives and distinguishes various face masks within 1/30th of a second—even within large crowd gatherings.

Privacy can be contained.

A system can be deployed rather fast.

The more deployments, the better. The more the AI watches, the smarter the AI gets. The more it sees, the more it "learns" to detect anomalies.

This tool can be applied to prevent active shooter events before they happen. Continuous monitoring and rapid identification could make a world of difference in getting first responders to a scene—when every second counts.

The Ph.D. expertise that built this

My company, Iterate.ai, is staffed with Ph.D.'s and creative thinkers who have high levels of AI expertise. We've built a dozen AI-based computer vision, edge computing, big data applications for commercial usage. We can deploy technology for social good, too. We believe that AI techniques can be applied to child safety, schools, churches, synagogues, mosques, courtrooms, and other public buildings.

Defining success

Success can be measured by counting alerts, reducing active shooter events, and getting first responders to crime scenes faster. Perhaps mental health metrics could be calculated, too. For instance, can we make students tomorrow feel safer than they do today? Can we decrease the number of violent events, which might also result in reducing anxiety?

Another success factor could be measured by the number of institutions that take advantage of this solution.

Ultimately, the end result will be safer environments and fewer injuries and deaths.

Next Step

If you know anyone who might find this information to be helpful, please forward this article or connect us: [email protected] or [email protected]