法藍克機電 Frank Mechatronics

Who says you need expensive hardware and powerful GPUs to build an industrial safety AI system?In this project, we challenge the limits of Edge Computing by building a fully functional HSE (Health, Safety, and Environment) Vision Alarm System running smoothly on a standard CPU! 5 Features of the System:

Written by

frankmec

—

×

Build an AI Virtual Fence (No GPU Required!)

Who says you need expensive hardware and powerful GPUs to build an industrial safety AI system?
In this project, we challenge the limits of Edge Computing by building a fully functional HSE (Health, Safety, and Environment) Vision Alarm System running smoothly on a standard CPU!

5 Features of the System:

  • Interactive Virtual Fence: Simply click 4 points on the live video feed to draw a custom, semi-transparent red alert zone (ROI).
  • Ultra-Fast Edge AI: Powered by the YOLO Nano model optimized for CPU, accurately tracking the position of human feet in real-time.
  • Smart Anti-False Alarm: Built-in 1-second delay mechanism. The alarm only triggers if a person stays in the danger zone , eliminating false positives!
  • Instant Audio & Visual Alerts: The system blasts a high-frequency warning beep, auto-saves a snapshot with bounding boxes, and updates the UI history list instantly.
  • Multi-Camera Smart Memory: Seamlessly switch between multiple USB/IP cameras. The system automatically saves and loads specific ROI coordinates for each individual camera!

Leave a comment