AI-Driven RF Sensing for Workplace Employee Health and Fitness Monitoring

Regular physical activity is vital for adults as it significantly contributes to overall health and well-being. To address the global challenge of physical inactivity, the goal is to achieve a 10% relative reduction by 2025 and a 15% reduction by 2030, compared to 2010 levels. Given that many adults spend 8 to 10 hours at their workplace, integrating exercises into the work environment provides an effective opportunity to promote physical activity. However, existing exercise monitoring systems, primarily camera-based, face challenges such as poor illumination and privacy concerns, limiting their suitability for workplace use. To overcome these limitations, this paper introduces the use of radar signals for monitoring employee fitness, with data represented as spectrograms. The system focuses on five exercise classes - Lower Back, Glutes, Arm Stretches, Neck Stretches, and Back Shoulder designed for individuals seated at their workplace. Deep learning (DL) models, including MobileNet, ResNet50, VGG16, and VGG19, process the radar data to classify exercise patterns. Among these, VGG16 demonstrates exceptional performance, achieving a classification accuracy of 100%, showcasing the effectiveness of radar-based monitoring for workplace fitness applications.