ESTIMATING THERMAL SENSATION OF OCCUPANTS IN INDOOR ENVIRONMENTS USING THERMAL IMAGE AND MACHINE LEARNING

The purpose of this study is to propose a machine-learning model that estimates the thermal sensation by utilizing thermal images and machine learning. Participants were tested in a laboratory, and a machine learning model was created for estimating the sensation from the thermal images using the training data obtained in the experiment. The model predicted three levels of occupants’ thermal sensation: “hot,” “neutral,” and “cold”. By group, the “cold group,” the “neutral group,” and the “hot group” were able to predict with 91.9%, 80.8%, and 95.4% accuracy, respectively, indicating that all groups were able to predict with high accuracy.