Measurement of Human Movement through Camera Motion Analysis based on Pose Estimation Algorithms: A Scoping Review
Shin-Woong Kim, Hae‐Rim Kim, Eun-A Shim, Kyung-Jae Lee, Byung-Mo Ku · Korean Journal of Sports Science · 2025
This study analyzes domestic research trends in camera-based motion analysis utilizing pose estimation algorithms, focusing on advancements in artificial intelligence and its applications in sports, rehabilitation, and posture analysis. The methodology involved a literature search, selecting 25 studies published between 2020 and 2024 based on recency, relevance, and publication in recognized journals. The findings reveal a growing trend in the adoption of camera-based motion analysis across multiple disciplines. In sports science, these technologies were used to enhance performance through motion tracking, while in rehabilitation, they provided continuous monitoring of patient progress and actionable feedback. One commonly used assessment tool, the TGMD-3, relies on subjective and labor-intensive assessments by evaluators. Advancements in algorithms such as MediaPipe, PoseNet, and OpenPose have the potential to address these limitations, offering more precise and efficient ways to assess human motion with reduced reliance on manual intervention. This review also identifies several challenges, including the need for higher algorithm reliability, standardization of evaluation methods, and broader applications beyond elite athletes to include vulnerable populations. Despite these limitations, the integration of human pose estimation with computer vision has significantly improved the objectivity and accessibility of motion analysis, offering potential improvements in athletic training, medical diagnostics, and ergonomic assessments. Future research should enhance algorithm validity and explore interdisciplinary applications to unlock their full potential in diverse fields.