A Real-Time Adaptive Thermal Comfort Model for Sustainable Energy in Interactive Smart Homes: Part II.

Clothing garments directly affect the human body's thermal balance and thermal comfort. The ideal thermal balance is when the body's temperature remains neutral and the environment is not affecting it. Nevertheless, achieving that thermal balance is very unlikely due to other variables, such as humidity, that need consideration. Therefore, these variables affect the human body's perception of the environment's temperature leading to behavioral problems and a lack of thermal comfort. Besides, adaptive methods require integrating dynamic models that predict clothing properties to provide accurate thermal comfort to the householder and understand how an individual adapts to indoor environments rather than the conventional thermal comfort analysis. Therefore, a computer vision system integrated into camera recognition is needed to implement an online clothing insulation recognition system to get feedback on thermal comfort and provide information to the householders about how the clothes and activities affect their thermal comfort. Besides, this recognition needs to be considered in dynamic interfaces such as connected thermostat interfaces. Furthermore, this vision system needs to detect the clothing worn by the user and infer possible metabolic activities based on the clothes. Hence, this paper proposes classifying the garments through a Deep Neural Network (DNN) using the YOLOv3 in which available external sources, such as cameras, gather the householder's clothes and postures to classify the type of cloth and activity and provide information to the householder through a dynamic interface in order to continue their thermal comfort. Thus, a 24-h simulation is performed considering three scenarios: (1) typical 0.5 clo value and 1.0 metabolic rate; (2) dynamic clo values with activities; and (3) dynamic values adding the underwear clo values. Hence, thermal comfort analysis results are included in an interactively connected thermostat mock-up. This mock-up and interaction are available online.