SENSING TECHNOLOGY OF PERSONAL HVAC USING HIGH-RESOLUTION INFRARED CAMERA

抄録

In this paper, we introduce a sensing technology to control personal HVAC (Heating Ventilation and Air Conditioning) system in office buildings. We use thermal images taken by a high-resolution infrared camera installed in an office to estimate indoor position of occupants, air temperature around the occupants and the clothes they are wearing. For the indoor position estimation, we calculate the distance between camera and person using the person’s head size, and obtain two-dimensional coordinates of the occupants. The air temperature is roughly estimated from the temperature distribution around the occupants in thermal images. For the clothes estimation, we classify occupant’s clothes into long-sleeves and short-sleeves categories by applying transfer learning based on YOLOv4.