基於動作影像追蹤於坐姿賦能運動量化評估與分析

目的：本研究目的為應用人工智慧動作影像追蹤技術量化分析坐姿運動器材之靜動態運動姿態及表現。方法：透過影像追蹤技術探討「主被動混合運動訓練器」採正確坐姿之適體擺位乘坐於可扭擺的座墊上，分別進行上肢平握及上肢上舉兩種姿態，並以兩種迴轉數 (60、80 RPM) 進行身體骨架節點的即時追蹤與計算，再以描述性統計和成對樣本 t 檢定進行驗證。結果：坐姿運動器材運作過程中，利用上下肢體肌肉的力量以維持身體之平衡確實在不同坐姿下的 60 與 80 RPM 達顯著差異。結論：基於本研究所開發的動作影像追蹤軟體平台，確實能提供人體動作的即時擷取和量化分析運動數據。本研究之整體架構與實驗設計皆呈現其可行性與實用性。  Purpose: The study aimed to apply artificial intelligence for motion capture and body tracking to analyze movement data characteristics of static and dynamic movement posture and performance. Methods: Using motion image tracking technology, the study employs a specially designed seat with adjustable rotation to ensure correct sitting posture, and the inventor of this study sits on this seat. Two upper limb positions, namely dual-hand grip and lifting movements, are performed at frequencies of 60 rpm and 80 rpm. Body skeletal joint calculations are conducted, and descriptive statistics and Paired Sample t-test are employed to validate the differences among the four combinations. Results: During the operation of the machine, utilizing the strength of the upper and lower limb muscles to maintain body balance indeed results in significant differences in the trajectory and characteristics of the four-movement groups. Conclusion: This research developed the image tracking software platform to validate and analyze quantitative movement data. The overall architecture and experimental design of the study are feasible and practicable.