EPPEC (Employee Personal Protection Equipment Checker)

EPPEC is an AI-powered system that verifies PPE in real time using camera detection, scenario training, and instant feedback for medical and disaster responders.

Comment

Inspiration

Emergency responders and healthcare workers operate in high-risk environments where proper Personal Protective Equipment (PPE) is critical. However, mistakes still happen—especially in high-stress scenarios like hurricanes, medical emergencies, or disaster response.

Training is often static, slow, and disconnected from real-world conditions. There is no immediate feedback loop to verify whether someone is properly equipped in real time.

We built EPPEC (Emergency PPE Compliance Checker) to solve this problem: → Provide real-time PPE verification + training using AI and computer vision.

What it does

EPPEC is an AI-powered system that ensures responders are wearing the correct PPE for a given scenario.

📡 System Flow

Camera Feed → YOLO PPE Detection → Confidence Scores → Scenario Rule Engine → Instant Feedback (Correct / Incomplete / Incorrect)

It supports two core modes:

🏥 Medical Mode

  • Select or generate clinical scenarios
  • Identify required PPE (Standard, Droplet, Airborne, etc.)
  • Verify correctness using AI + rule-based grading

🌊 Hurricane / Disaster Mode

  • Ensures responders wear appropriate protective gear
  • Alerts those in view to leave if they do not have a hard hat on

📸 Core Features

  • Live camera PPE detection
  • Confidence scoring for each PPE item
  • Scenario-based training workflow
  • Instant feedback (correct / incomplete / incorrect)

How we built it

EPPEC is a full-stack application combining computer vision, AI, and web technologies.

Backend

  • Built with FastAPI to handle classification, detection, and grading
  • Implements GeminiAPI for generation of alternative scenarios
  • REST endpoints for scenarios, detection, and submissions

Computer Vision

  • Real-time PPE detection using a YOLO11n model
  • Produces bounding boxes, class labels, and confidence scores per frame
  • Achieves ~0.87–0.90 mAP50 and ~0.59–0.61 mAP50-95 on validation data
  • Runs with sub-second latency for live camera feedback
  • Supports multiple PPE classes including masks, gloves, goggles, and coveralls

AI Integration

  • Generates realistic training scenarios dynamically
  • Classifies PPE requirements based on scenario text

Frontend

  • Built with React + TypeScript + Vite with TailwindCSS
  • Live camera feed + real-time confidence indicators
  • Interactive PPE selection and feedback UI

Challenges we ran into

  • Balancing model size (YOLO11n vs YOLO11s) to maintain real-time performance while improving detection accuracy
  • Normalizing inconsistent labels from datasets for the vision model
  • Synchronizing live camera input with backend processing
  • Designing an intuitive UI for high-pressure environments

Accomplishments that we're proud of

  • Built a fully functional end-to-end AI detection system in a short timeframe
  • Achieved real-time PPE detection with high confidence scoring
  • Integrated AI-generated scenarios into training workflows
  • Created a system usable in both medical and disaster response settings

What we learned

  • How to integrate computer vision into web applications
  • Designing real-time feedback systems
  • Combining AI models with rule-based validation
  • Building scalable training tools for safety-critical environments

What's next for EPPEC (Employee Personal Protection Equipment Checker)

  • Improve detection accuracy with larger datasets
  • Deploy on mobile devices for field use
  • Expand to other domains (firefighters, industrial safety, military)
  • Integrate with real-world emergency response systems

Why this matters...

EPPEC reduces human error in safety-critical environments by introducing an automated, real-time feedback loop, something traditional training systems lack.

In high-pressure scenarios like medical emergencies or disaster responses, even a single missing piece of PPE can lead to serious consequences. EPPEC ensures responders receive immediate, objective feedback, helping prevent mistakes before they happen, not after.

References

Built With

Share this project:

Updates