人工智慧已成為解決各領域問題的工具，而在人體動作辨識技術快速發展的時代，其應用領域已十分廣泛。若可以將此技術應用於辨識現場人員作業的步驟，將有效改善人員的操作績效與安全。本研究使用人體動作辨識方法建立作業步驟分類模型，並觀察其分類模型性能，以協助工業工程師或現場管理人員了解現場作業績效。本研究設計五組人工物料作業實驗分別為組裝小紙箱、組裝大紙箱、裝箱作業、組裝小木箱及組裝大木箱作業並制定作業的標準步驟，以模擬作業人員的操作，本研究利用深度攝影機(Kinect)提取人體的骨骼資料，使用非監督式學習應用於步驟分群，監督式分類演算法建立分類模型。並以運動學特徵擷取方法及動態時間扭曲(Dynamic Time Warping, DTW)處理時間序列資料。研究結果顯示單純使用原始資料分群及分類效果不盡理想，在本研究中以運動學特徵資料擷取以演算法（隨機森林、樸素貝葉斯、支持向量機、K-近鄰）進行分類以及動態時間扭曲結合K-近鄰演算法進行分類皆可以準確地分類作業的步驟。結果顯示，在工作場域中使用機器學習技術辨識人體動作具備高度可行性。

Artificial intelligence has become a tool to solve problems in various fields. In the era of rapid development of human action recognition technology, the application of it has already been deployed widely. If it can be applied to identify the steps of on-site personnel operations, it will effectively improve the performance and the safety of the personnel. In this study, the human action recognition method was used to establish a classification model of different activities. Several machine learning techniques were introduced to analyze the performance of the classification model, which can be further used to assist industrial engineers or site managers to understand the performance of on-site operations. The experiment includes five groups of operators performing different activities, namely, assembling small cartons, assembling large cartons, packing operations, assembling small wooden boxes, and assembling large wooden boxes. By using Kinect to extract the skeletal motion data from the operators, this study applies unsupervised learning to clustering steps and uses supervised classification algorithms to build classification models. In addition, the study uses kinematic feature extraction methods and dynamic time warping (DTW) to process time series data. The results of the study showed that the performance of the clustering and classification model created by raw data alone is not ideal. However, using kinematic feature extraction and the algorithm, such as Random Forest, Naive Bayesian, Support Vector Machine, K-nearest neighbors, and Dynamic Time Warping combined with K-nearest neighbors, the accuracy of the identification process is significantly improved. These result shows the feasibility of the application of machine learning technique to identify human motions.

Afrasiabi, M., & Mansoorizadeh, M. (2020). DTW-CNN: time series-based human interaction prediction in videos using CNN-extracted features. The Visual Computer, 36(6), 1127-1139.
Bagnall, A., Bostrom, A., Large, J., & Lines, J. (2016). The great time series classification bake off: An experimental evaluation of recently proposed algorithms. Extended version. arXiv 2016. arXiv preprint arXiv:1602.01711.
Ball, A., Rye, D., Ramos, F., & Velonaki, M. (2011). A comparison of unsupervised learning algorithms for gesture clustering. Paper presented at the 2011 6th ACM/IEEE International Conference on Human-Robot Interaction (HRI).
Ball, A., Rye, D., Ramos, F., & Velonaki, M. (2012). Unsupervised clustering of people from'skeleton'data. Paper presented at the Proceedings of the seventh annual ACM/IEEE international conference on Human-Robot Interaction.
Breiman, L. (2001). Random forests. Machine learning, 45(1), 5-32.
Cho, T. Z. W., Win, M. T., & Win, A. (2018). Human Action Recognition System based on Skeleton Data. Paper presented at the 2018 IEEE International Conference on Agents (ICA).
Clark, R. A., Pua, Y.-H., Fortin, K., Ritchie, C., Webster, K. E., Denehy, L., & Bryant, A. L. (2012). Validity of the Microsoft Kinect for assessment of postural control. Gait & posture, 36(3), 372-377.
Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine learning, 20(3), 273-297.
Cover, T., & Hart, P. (1967). Nearest neighbor pattern classification. IEEE transactions on information theory, 13(1), 21-27.
Dranca, L., de Abetxuko Ruiz de Mendarozketa, L., Goñi, A., Illarramendi, A., Navalpotro Gomez, I., Delgado Alvarado, M., & Rodríguez-Oroz, M. C. (2018). Using Kinect to classify Parkinson’s disease stages related to severity of gait impairment. BMC bioinformatics, 19(1), 1-15.
Ignatov, A. (2018). Real-time human activity recognition from accelerometer data using Convolutional Neural Networks. Applied Soft Computing, 62, 915-922.
Janidarmian, M., Roshan Fekr, A., Radecka, K., Zilic, Z., & Ross, L. (2015). Analysis of motion patterns for recognition of human activities. Paper presented at the Proceedings of the 5th EAI International Conference on Wireless Mobile Communication and Healthcare.
Kong, W., He, L., & Wang, H. (2021). Exploratory data analysis of human activity recognition based on smart phone. IEEE Access, 9, 73355-73364.
Kotsiantis, S. B., Zaharakis, I. D., & Pintelas, P. E. (2006). Machine learning: a review of classification and combining techniques. Artificial Intelligence Review, 26(3), 159-190.
Leung, K. M. (2007). Naive bayesian classifier. Polytechnic University Department of Computer Science/Finance and Risk Engineering, 2007, 123-156.
Liu, F., Zeng, W., Yuan, C., Wang, Q., & Wang, Y. (2019). Kinect-based hand gesture recognition using trajectory information, hand motion dynamics and neural networks. Artificial Intelligence Review, 52(1), 563-583.
Lv, W. (2021). Gesture recognition in somatosensory game via Kinect sensor. Internet Technology Letters, e311.
MacQueen, J. (1967). Classification and analysis of multivariate observations. Paper presented at the 5th Berkeley Symp. Math. Statist. Probability.
Maharani, D. A., Fakhrurroja, H., & Machbub, C. (2018). Hand gesture recognition using k-means clustering and support vector machine. Paper presented at the 2018 IEEE Symposium on Computer Applications & Industrial Electronics (ISCAIE).
Marcacini, R. M., Carnevali, J. C., & Domingos, J. (2016). On combining websensors and dtw distance for knn time series forecasting. Paper presented at the 2016 23rd International Conference on Pattern Recognition (ICPR).
Miao, S., Shen, C., Feng, X., Zhu, Q., Shorfuzzaman, M., & Lv, Z. (2021). Upper limb rehabilitation system for stroke survivors based on multi-modal sensors and machine learning. IEEE Access, 9, 30283-30291.
Plantard, P., Muller, A., Pontonnier, C., Dumont, G., Shum, H. P., & Multon, F. (2017). Inverse dynamics based on occlusion-resistant Kinect data: Is it usable for ergonomics? International Journal of Industrial Ergonomics, 61, 71-80.
Priya, S. J., Rani, A. J., Subathra, M., Mohammed, M. A., Damaševičius, R., & Ubendran, N. (2021). Local pattern transformation based feature extraction for recognition of Parkinson’s disease based on gait signals. Diagnostics, 11(8), 1395.
Rabiner, L., & Juang, B.-H. (1993). Fundamentals of speech recognition: Prentice-Hall, Inc.
Rahman, M. W., & Gavrilova, M. L. (2017). Kinect gait skeletal joint feature-based person identification. Paper presented at the 2017 IEEE 16th international conference on cognitive informatics & Cognitive Computing (ICCI* CC).
Shotton, J., Fitzgibbon, A., Cook, M., Sharp, T., Finocchio, M., Moore, R., . . . Blake, A. (2011). Real-time human pose recognition in parts from single depth images. Paper presented at the Computer Vision and Pattern Recognition (CVPR), 2011 IEEE Conference on.
Shotton, J., Sharp, T., Kipman, A., Fitzgibbon, A., Finocchio, M., Blake, A., . . . Moore, R. (2013). Real-time human pose recognition in parts from single depth images. Communications of the ACM, 56(1), 116-124.
Sundstrup, E., Seeberg, K. G. V., Bengtsen, E., & Andersen, L. L. (2020). A systematic review of workplace interventions to rehabilitate musculoskeletal disorders among employees with physical demanding work. Journal of occupational rehabilitation, 30(4), 588-612.
Wan, X., Li, H., Zhang, L., & Wu, Y. J. (2021). Multivariate Time Series Data Clustering Method Based on Dynamic Time Warping and Affinity Propagation. Wireless Communications and Mobile Computing, 2021.
Wang, S., Yang, J., Chen, N., Chen, X., & Zhang, Q. (2005). Human activity recognition with user-free accelerometers in the sensor networks. Paper presented at the Neural Networks and Brain, 2005. ICNN&B'05. International Conference on.
Widodo, A., Yang, B.-S. J. M. s., & processing, s. (2007). Support vector machine in machine condition monitoring and fault diagnosis. 21(6), 2560-2574.
Wu, W. H. (2005). Medical Embedded Device for Individualized Care (MEDIC). University of California, Los Angeles,
Yadav, S. K., Tiwari, K., Pandey, H. M., & Akbar, S. A. (2021). A review of multimodal human activity recognition with special emphasis on classification, applications, challenges and future directions. Knowledge-Based Systems, 223, 106970.
Yang, G.-Z., & Yang, G. (2006). Body sensor networks (Vol. 1): Springer.
Yu, X., & Xiong, S. (2019). A dynamic time warping based algorithm to evaluate kinect-enabled home-based physical rehabilitation exercises for older people. Sensors, 19(13), 2882.
Zhang, C., & Lu, Y. (2021). Study on artificial intelligence: The state of the art and future prospects. Journal of Industrial Information Integration, 23, 100224.
Zhang, H., He, X., & Liu, Y. (2020). A Human Skeleton Data Optimization Algorithm for Multi-Kinect. Paper presented at the 2020 Asia-Pacific Conference on Image Processing, Electronics and Computers (IPEC).