Visio's Fire Detection On The Edge in Real-Time

Visio's Fire Detection On The Edge in Real-Time

Visio's Fire Detection On The Edge is an image processing based solution to detect fire and smoke using standard CCTV cameras. The detection software can identify fire as small as 30 pixels by 30 pixels (px) at more than a 99% accuracy rate, with the fire’s precise location (latitude and longitude) appearing on a calibrated camera in real-time as well as predicting the possible fire intensity. For reference, this means that for a frame resolution of 1080p, 30 px X 30 px will be around 4 percent of the frame. The detection software can work in a variety of environments – even those with poor lighting, challenging backgrounds, and inclement weather conditions. Possible applications include fire detection and alerts in hospitals, schools, commercial kitchens, offices, national infrastructures, zoos, retail premises and manufacturing plants. Other novel use cases being investigated include hazard alerts for oil and gas industries, fire and smoke detection for forest fires.

*Please note that member solutions are often customizable to meet the needs of individual enterprise end users.



  • Accuracy
    More than a 99% accuracy rate in detecting even 4% of the frame
  • Predictability in real-time
    Predicting the possibility of fire intensity in real-time


Worldwide Intel® Movidius™ Intel® Neural Compute Stick (NCS2) On-premise (Private Cloud, Other) Caffe TensorFlow Automotive Cross-Industry Energy and Utilities Manufacturing Models cant be re-trained - Inference only AI Software / SaaS Linux Intel® Distribution for Python Intel® Movidius™ Myriad™ Development Kit (MDK) Intel® Movidius™ Neural Compute SDK Yolo Image / Object Detection / Recognition / Classification Video Surveillance and Analytics Deep Learning