Abstract
As technologies continue to grow every day, more people are adapting to sedentary lifestyles, as they spend long hours sitting at their workspace during their workday. This sitting behavior is linked to many health issues, including a variety of musculoskeletal pains, which affect employees' well-being and result in high healthcare costs for employers. Various methods have been proposed before for human body landmark detection, but they tend to be more suitable for a more complex system design. This thesis introduces a more straightforward solution using OpenCV, an opensource computer vision library, and MediaPipe Pose, a human body landmark estimation developed by Google, to address the challenges of posture detection. This paper presents a method consisting of a non-wearable and smart automatic wellness device that's designed to promote healthy sitting habits by integrating computer vision technology into everyday life activities. The paper discusses different approaches and testing methods to develop an effective communication system between the hardware and software user interface. The paper also looks at the implementation of Bluetooth Low Energy (BLE) communication for a user-friendly and seamless interaction, offering real-time feedback to users about their overall sitting posture. Furthermore, the system efficiently stores user data collected from sensors for long-term analysis of posture trends and improvements.