Developing a taxonomy for vision-based personal thermal comfort: A guiding systematic review for detection, prediction, and control

Personal Thermal Comfort (PTC) provides data-driven control for Heating, Ventilation, and Air Conditioning (HVAC) systems. Conventional invasive assessment methods retain real-world implementation. Vision-based PTC studies provide methodological variations, but there is a lack of a comparable and unified modelling and assessment framework. This systematic review synthesises literature aiming to develop a vision-based PTC taxonomy and a modelling framework, highlighting methodological strengths and weaknesses. The PRISMA framework was utilised to analyse 38 studies from Scopus and Web of Science with predefined inclusion criteria. The taxonomy emphasises a focus on comfort-based rather than demand-based predictions. Among key features, image-extracted facial temperature achieved an 85% accuracy in predicting thermal sensation. Ensemble models showed high prediction consistency with 85% accuracy. Gaps remain in investigating gender, climatic, and operational variations. The framework is a guiding protocol for researchers and practitioners seeking real-time, non-invasive, occupant-centred comfort and energy optimisation. It serves as a decision manual for comfort personalisation and life-cycle sustainability assessment.