Gas sensors and real-time video for accurate classroom occupancy detection

Area of Science:

• Building Science
• Computer Vision
• Environmental Sensing

Background:

• HVAC systems consume significant energy, with efficiency often compromised by inaccurate occupancy data.
• Optimizing HVAC operations based on real-time occupancy is crucial for energy conservation in smart buildings.

Purpose of the Study:

• To develop and validate a dual-modal system for accurate, real-time classroom occupancy detection.
• To enhance HVAC efficiency and building sustainability through precise occupancy monitoring.

Main Methods:

• Integration of video feeds (Logitech C20 webcam) and gas sensor data (MS1100).
• Utilized YOLOv4 object detection model trained on over 20,000 labeled human face images.
• Employed OpenCV for real-time video stream processing and analysis.

Main Results:

• Achieved over 98% accuracy in real-time occupant identification and counting.
• Demonstrated improved detection reliability under diverse environmental conditions by combining sensor modalities.
• Validated a robust and scalable framework for dynamic HVAC control.

Conclusions:

• The proposed dual-modal system offers a significant advancement in real-time occupancy detection for HVAC optimization.
• This methodology presents a practical and scalable solution for reducing energy consumption in smart buildings.
• Enhanced accuracy and reliability in occupancy detection contribute to improved building sustainability and energy efficiency.