在精細動作發展的應用中，應考慮即早性、普遍性、訓練的持續性和運動過程的紀錄。許多研究在兒童時期即給予早期評估和療育，然而並非所有的精細動作發展遲緩患者都能得到良好的療育資源，傳統的介入需要治療師在場並使用訓練系統和評估量表進行療育，在目前供不應求的情況下，治療師自然難以到偏鄉給予長期的治療。因此，我們開發了一款線評估的遠端系統，並提出了一種新穎的精細動作品質評估方法，該方法通過使用人機互動、骨架追蹤技術收集兒童的精細運動，然後使用深度學習模型對兒童的精細動作質量進行分析。使用人機交互技術來開發遊戲可以提高孩子的參與感並記錄整個評估過程，並搭配Leap Motion Controller (LMC)感測器，它具有出色的手部追蹤效果。為了證明分層共現網絡識別動態精細動作的有效性，使用Shape Retrieval Contest（SHREC）數據集進行驗證，結果也證明此方法的準確性勝過當前文獻的方法。之後，我們使用轉移學習技術來轉移公開數據集的知識，實驗也證實了更好的評估。;In the application of the development of fine motor, consideration should be given to the earlyness, universality, continuity of training and the record of the motor process. Many research have been given early assessment and intervention during childhood, but not all patients with developmental delays of fine motor can receive good treatment resources. Traditional intervention requires therapist to use the training system and assessment scale for treatment on the spot, which is currently in short supply. it is difficult for the therapist to go to a remote area for long-term treatment. Therefore, we have developed a remote system that can achieve online assessment, and also proposed a novel method for fine motor quality assessment that collect children′s fine motor using human–computer interaction, skeleton tracking technology, and then use the deep learning model to analyze the quality of the fine motor. Using human-computer interaction technology to develop games can increase children′s sense of participation and record the entire assessment process. Hand interaction uses Leap Motion Controller (LMC), which has excellent hand tracking effect. In order to prove the effectiveness of the Hierarchical Co-occurrence Network (HCN) to recognize dynamic fine motor. we used the Shape Retrieval Contest (SHREC) dataset for verification, the method was also proven to outperform in accuracy competing approaches of the current literature. After that, we use transfer learning technology to transfer the knowledge of the open dataset, the experiment also confirmed a better assessment.
