Thermal Comfort Evaluation Method Using Facial Thermal Imaging

Indoor comfort profoundly impacts productivity and health. Current evaluation methods rely on subjective questionnaires and physiological measurements, offering both subjective and objective insights. However, questionnaires can be influenced by varying user perceptions, suitable mainly for lab settings. Some physiological measurement tools are intrusive and less practical in certain situations. Furthermore, physiological changes may not promptly align with perception changes, leading to a lag. Recognizing that facial skin temperature varies with thermal conditions, this paper proposes a thermal comfort evaluation method using facial thermal imaging. It begins with non-contact thermal imaging for facial images. A deep learning model is then built to learn temperature patterns under diverse conditions. This model accurately assesses thermal comfort, with a 97.7% accuracy rate compared to questionnaires. Thus, facial thermal imaging proves effective for thermal comfort evaluation in varied conditions.