# AI Violence Detection for K-12 Schools

> A real-time, privacy-first computer vision system that turns the IP cameras you already have into a violence-detection layer — and notifies the right adult before a fight escalates.

## The problem most schools face today
Every K-12 school in the country already has cameras. Most don't have anyone watching them. By the time staff sees a fight on replay, it's on TikTok, someone's in the ER, and parents are calling the principal. Traditional CCTV is a forensic tool: it tells you what happened *after* it happened.

GuardianAI converts that same camera feed into a real-time alerting system. A pose-based AI model watches every stream 24/7 and pings a school resource officer (SRO), vice principal, or designated on-call adult when it sees fighting, hitting, or shoving — usually within 1–2 seconds of the first contact.

## How it works in a school environment
1. **Connect existing cameras.** Any RTSP stream works — Hikvision, Dahua, Avigilon, Axis, Reolink, even a USB webcam. No new hardware required.
2. **On-prem AI inference.** The detection runs on a single mid-tier GPU box (NVIDIA Jetson Orin or equivalent) installed inside the school's network. Video never leaves the building.
3. **Skeleton-only analysis.** YOLO11n-Pose extracts 17 keypoints per person, then a CTR-GCN spatiotemporal graph network classifies motion. No faces. Just geometry.
4. **Instant Telegram alert.** When the model fires, the on-call adult gets a message with the camera location, a confidence score, and a short clip preview. They confirm or dismiss with one tap.

## Why pose-based, not face recognition
- **Legal.** Face recognition for K-12 surveillance is illegal in many U.S. states (e.g., New York's 2023 moratorium), violates GDPR Article 9 in the EU, and creates a permanent biometric record of minors.
- **Architectural privacy.** The CTR-GCN classifier looks at how bodies move relative to each other — not who they belong to. The model literally cannot identify a student.
- **Bias-resistant.** Skeletons strip away skin tone, clothing, and lighting. Removes the most common sources of fairness failures in pixel-based surveillance AI.

## What schools get out of the box
- **Live dashboard** — single web view showing every connected camera, with red-bordered overlays when violence is detected.
- **Telegram bot** — per-school chat groups for SROs, admins, counselors. Each alert includes one-tap "confirm" / "false alarm" feedback.
- **Incident log** — searchable history with timestamps and resolution labels — exportable for board reports.
- **Analytics** — heat-map of incident times and locations across the school day; informs supervision adjustments.

## Deployment timeline
A typical 800-student building goes from contract signature to first live alert in **10–14 days**. Bulk of that time: network configuration (RTSP credentials per camera) and one site visit to install the on-prem GPU box.

## Pricing model
Annual per-camera subscription that includes the on-prem appliance, software updates, model retraining, and 24/7 health monitoring. No per-incident or per-alert fees. Quote: hello@guardianai.tech.

## Frequently asked questions

### Does GuardianAI work on existing analog cameras?
Yes, as long as the camera feeds into a DVR or NVR that exposes RTSP. About 95% of units installed in U.S. schools since 2018 do.

### Can it run if our internet goes down?
Yes. Detection is fully local. Dashboard continues on the school LAN; only Telegram alert delivery is interrupted, and queued messages flush when connectivity returns.

### What about FERPA, COPPA, GDPR?
The system never stores or transmits identifiable images of minors, so it sidesteps the disclosure rules that apply to traditional CCTV. We provide a Data Processing Agreement and DPIA template on request. See https://guardianai.tech/privacy.

### Will it generate false alarms?
Yes — every detection model does. Reported precision on a held-out benchmark of school CCTV clips is 0.91 with a 0.5 confidence threshold. The confirm/false-alarm feedback loop reduces noise per-school over the first 90 days.

### Is it appropriate for elementary schools?
Yes. Trained partly on UBI-Fights and curated school cafeteria/playground clips. We recommend a higher threshold (0.65+) in elementary settings to filter out roughhousing.

## Related reading
- [University campus deployments](https://guardianai.tech/use-cases/campuses/index.md)
- [How pose detection works](https://guardianai.tech/technology/pose-detection/index.md)
- [How the spatiotemporal classifier works](https://guardianai.tech/technology/spatiotemporal-graph/index.md)

---
*Markdown mirror of https://guardianai.tech/use-cases/schools.*