In our recent article, Visual AI for School Safety: Time to Act, we described the devastating impact of school shootings on our communities and the urgent question K-12 and higher education leaders are grappling with today: what more can be done to safeguard our students, teachers, and staff from active shooter threats?
It’s both a question and a call to action: what more can we do? A starting point is to assess what we are currently doing, what we should continue doing, and what we should begin doing to improve school safety.
What are we currently doing to safeguard schools?
There are over 130,000 K-12 public and private schools in the U.S. of all shapes and sizes. Different schools have different policies, funding, settings, circumstances, etc., but the common goal of all schools is to teach students. It isn’t to lock them up in an impenetrable vault—schools are learning environments that need to balance normalcy as well as security.
Over the past two decades, school systems have enacted sharp increases in security measures to try to “harden” their facilities from active shooter threats. A 2019 National Center for Education Statistics study found that two out of three schools (and over 80% of middle and high schools) had placed at least one school security or police officer on campus. The NCES found that almost all schools:
- Controlled access to their building during school hours
- Used security cameras
- Planned for potential active shooter events and conducted lockdown drills.
Other safety and security measures include personal identification/badge systems, metal detectors, gunshot detectors, bullet-proof backpacks, and random security sweeps.
All of those steps can be important elements in a school’s security protocol. However, none can detect and prevent active shooter incidents alone, with each serving a limited function that could potentially be circumvented. Combined, they don’t add up to a scalable solution that most school districts can effectively acquire and manage. And many of these measures provide their value reactively—after an incident is too far in progress to avert tragic consequences.
For instance, most schools currently have surveillance cameras, but how are they being used to detect and prevent school shootings? Their focus leans toward post-incident investigations, i.e., something happens, then authorities review the video footage to reconstruct the timeline and document evidence. They are not likely to be monitored in real time, nor at all times, because this is a tedious, inefficient, and expensive operating model. Studies have shown that humans performing the task of watching video feeds struggle to pay attention for more than 20 minutes. Engagement plummets rapidly when more than one camera competes for attention.
Another example is gunshot detection technology. While the ability to audibly detect gunshots offers widespread coverage, it is essentially a reactive measure that occurs after the first shot, when lives are already at risk. It also doesn’t provide effective location accuracy capability to first responders, so police and medical teams must allocate crucial minutes assembling this information themselves instead of rendering aid as soon as possible during a crisis.
Acoustic or radar-enabled concealed weapon detectors can discover weapons in specific locations (e.g., a metal detector in an entryway), which is undoubtedly a helpful prevention and detection measure. But this solution is limited in range and coverage area, usually a fixed location. Active shooters can evade these areas, entering through an unguarded door.
What current security measures should we continue using to safeguard schools?
If current measures are flawed or incomplete, that doesn’t mean they don’t still serve a purpose. Decision-makers should carefully evaluate their exposure and mitigation strategies and try to address as many security gaps as possible using all the tools they consider valid. Layers of security, even with some redundancy or imperfection, are expected. But a disciplined approach would replace or improve measures that only deliver a false sense of security or so-called ‘security theater.’ The stakes are too high to allocate funding or trust in measures that provide negligible value. And it is also important to prioritize security measures that provide the greatest capabilities with the least interference in students’ and teachers’ daily lives.
What more can we do to improve school safety?
Advances in artificial intelligence, especially computer vision, are driving a paradigm shift in how we use camera systems for safety, security, and other high-value use cases. Solutions leveraging visual AI are proliferating rapidly today in commercial applications as well as the public sector. And since the majority of school campuses already utilize surveillance cameras, an opportunity arises: what if schools could transform their underutilized video feeds into a dynamic, proactive visual AI platform—using computer vision technology to identify and classify objects and determine the next best actions based on what their cameras “see?”
Our product, SparkCognition Visual AI Advisor, integrates rapidly with existing closed-circuit cameras to detect, alert, track and prevent security threats in real time, 24/7. Wherever cameras can see inside or outside of school buildings, Visual AI Advisor can autonomously analyze what is happening and respond accordingly. When it detects behavior indicative of an active shooter situation—whether it’s a person with a gun, someone needing first aid, a door propped open suspiciously, or many other dynamic scenarios—it will warn safety personnel while maintaining vigilance as the situation develops. Visual AI Advisor automatically sends role-based alerts to notify designated personnel of the threat with vital details (e.g., location, time, and visual ID) via mobile apps, email, and text and can tie into a school’s onsite security system—a critical advantage when seconds count.
Visual AI Advisor can run dozens of critical security use cases on a single camera, drawing from a portfolio of over 120 pre-built machine learning models to take school safety to a new standard. It includes everything from situational awareness of suspicious behavior to detecting individuals entering unauthorized areas or identifying unfamiliar vehicles in the school parking lot. It’s a highly scalable solution for schools with favorable set-up and operating costs compared to prevalent security measures in use today. Visual AI Advisor deploys rapidly to your existing camera infrastructure with a low-code/no-code integration framework. Privacy is built into Visual AI Advisor by design, deploying securely on-prem or in the cloud without the need to store video files, plus anonymization features to comply with a school’s privacy policies and practices.
SparkCognition Visual AI Advisor is a highly versatile, intelligent, always-on solution leveraging the power of visual AI to transform school safety from a reactive approach to a proactive one. It provides a compelling answer to the question on so many minds today—what more can we do to safeguard schools?—by scaling real-time automatic surveillance and alerting capacity across your entire campus.