COMPUTER VISION-BASED OCCUPANCY DETECTION SYSTEMS AND APPROACHES: A REVIEW

Abstract

Occupants are the core of the built environment. Developments in information technology facilitate gathering data on occupant presence and behaviour and learning from it through data-driven methods. Recent studies highlight the effectiveness of integrating occupancy data into HVAC control systems to reduce energy consumption while maintaining occupants’ thermal comfort. This paper presents a comprehensive review of computer vision-based occupancy detection systems and approaches, mainly applied in the domain of controlling thermal environment. A systematic literature review methodology was employed to provide a structured process for the selection and analysis of studies. The paper aims to define occupancy parameters critical to occupant-centric HVAC control and personalized thermal comfort. It explores how computer vision techniques detect these parameters, with a focus on tools, deep learning models, dataset sizes, and evaluation metrics. Moreover, it summarizes various approaches that can be implemented for real-time indoor occupancy detection. The review analyses 43 papers published over the past decade. Findings emphasize the role of machine learning and computer vision in developing intelligent, occupant-centric control systems that optimize both comfort and energy efficiency. Finally, current challenges and future directions are discussed.