Design and Evaluation of an Online Squat Fitness System: Lessons Learned During the Early COVID-19 Pandemic in Japan

Abstract

COVID-19 has changed our lives and limited our ability to have adequate physical activity (PA). It is necessary to replace outdoor PA with home-based fitness. However, people lack access, skills, and even motivation for home-based fitness. To address these issues, we designed a free access self-monitoring and coaching and music-based interactive online squat fitness system. Body weight squat was utilized for fitness exercise and evaluated based on three indices: knee width, hip depth, and rhythm. An online survey on changes in exercise due to the COVID-19 pandemic and exercise habits was conducted to investigate the effect of the COVID-19 pandemic on PA. We collected data from 557 respondents 5 months after the system first released and analyzed 200 visitors' performance on squat exercise and the other relevant parameters. Visitors were divided into three groups according to their age: younger, middle, and older groups. Results showed that the younger group had better squat performance than the middle and older groups in terms of hip depth and rhythm. We highlighted the lessons learned about the system design, fitness performance evaluation, and social aspects, for future study of the design and development of similar home-based fitness systems. We provided first-hand results on the relation between the COVID-19 pandemic and physical exercise among different age groups in Japan, which was valuable for policy making in the post-COVID-19 era.

The Effect of Online Low-intensity Exercise Training on Fitness and Cardiovascular Parameters

“…The subjects in that study performed 20 exercise sessions by following the device guide on a laptop and reported that those exercises improved the physical and mental well-being; however, there was no control group. Further, Wang et al [27] reported regarding an original online exercise evaluation system, which effectively estimated the techniques of squat exercise in young, middle-aged, and older populations. Overall, such technologies may motivate general populations and give rise to meaningful health-oriented content within online exercise programs not only for young populations but also for the middle-aged and older populations.…”

Front-end deep learning web apps development and deployment: a review

“…MoveNet is a lightning-fast and highly accurate model that detects the body’s 17 key points; BlazePose can detect 33 keypoints, and PoseNet can detect multiple poses, each of which includes 17 key points. Works on Pose-detection can be found in [ 20 , 32 , 81 , 85 , 92 , 96 , 96 , 98 , 106 , 116 , 124 ] BodyPix This is a body segmentation model that segments 24 body components from a background image or video in real-time, and it also works for multiple people. Its design is based on either MobileNetV1, a smaller but less precise model, or ResNet50, a bigger but more precise model.…”

“…According to them, a single RGB camera may miss certain essential information, and incorrect body movement during workouts may also be overlooked. A self-monitoring and coaching system for online squat fitness is also designed in [ 124 ].…”