Development of a Deep Neural Network Model for Estimating Joint Location of Occupant Indoor Activities for Providing Thermal Comfort

The type of occupant activities is a significantly important factor to determine indoor thermal comfort; thus, an accurate method estimate activity needs be developed. purpose this study was develop deep neural network (DNN) model for estimating the joint location diverse human activities, which will used provide comfortable environment. DNN trained with images 14 joints person performing 10 common activities. contained numerous shortcut connections efficient training and had two stages sequential parallel layers localization. Estimation accuracy quantified using mean squared error (MSE) estimated percentage correct parts (PCP) body parts. results show that MSEs head neck were lowest, PCP highest torso. individual ranged from 0.71 0.92, while typing standing in relaxed manner PCP. higher relatively still lower involving wide-ranging arm or leg motion. This thus highlights potential estimation by proposing novel model. approach holds significant promise finding actual use target applications related comfort buildings.