近年來健康意識抬頭，人們越來越關心自身的健康狀況，而規律的運動是維持健康最好的方法。在各項運動中游泳為全身性的運動，也是少數在人生各種階段皆可從事的運動。為了量化運動成果，一套能在游泳時記錄划水次數、趟數與泳姿辨識等資訊的系統，不僅能讓專業的運動員提升自身的游泳技術，也能讓一般使用者記錄每次游泳的資訊。
本研究利用智慧型手錶內建的加速度計與陀螺儀來擷取使用者划水時的動作訊號，並開發一游泳資訊分析的演算法，提供使用者划水次數、趟數與泳姿辨識等資訊。演算法分為兩部分，訊號預處理階段與划水資訊分析階段。訊號預處理主要是為了排除移動雜訊的問題，以提高划水計數與趟數的準確率；划水資訊分析階段為划水次數計數、趟數計數與泳姿辨識。
實驗結果顯示，本研究在划水次數、游泳趟數上的平均誤差分別為1.58±1.79次、0.19±0.39趟，泳姿辨識的平均準確率為97.5%。與目前市售產品相比，本研究提出的方法能較準確的計算划水次數、游泳趟數，以及準確的辨識泳姿。

With the health consciousness growing in recent years has led many people to increasingly concerned their own health. However, regular exercise is the best way to stay healthy. Swimming is a full-body workout which is one of the few exercises that can be done through all stages of life. In order to quantify the results of sports. A system which can record stroke, lap and even swimming style automatically can't only assist professional swimmers to improve their swimming, but also can convenient to casual swimmers record their exercises results.
In this study, we use accelerometer and gyroscope, which is built in the smartwatch capture the activity signal during swimming and developing an algorithm for the analysis of the swimming information. It provides information on stroke count, stroke lap, and swimming style identification. Our algorithm consists of two parts, signal pre-processing phase and stroke analysis phase. In order to improve the accuracy of stroke count and lap count, signal pre-processing is mainly to eliminate the noise. The stroke analysis phase is for stroke count, lap count and identification of swimming styles.
The experiment shows that the mean error of our algorithm in stroke counts is1.58±1.79 times, in lap counts is 0.19±0.39 lap and the accuracy of swimming styles identification is 97.5%. Comparing with the current commercial product, the algorithm proposed by this paper, can effectively calculate the stroke count, lap count and identification of swimming styles.

[1] Carolyn Crist, “Swim fitness trackers work well enough for recreational swimmers.” [Online]. Available: https://www.reuters.com/article/us-health-swimming-fitness-trackers/swim-fitness-trackers-work-well-enough-for-recreational-swimmers-idUSKBN15Z1XF.
[2] BJSM, “Major new study on health benefits of swimming released - BJSM blog - social media’s leading SEM voice,” 2017. [Online]. Available: http://blogs.bmj.com/bjsm/2017/06/23/major-new-study-health-benefits-swimming-released/.
[3] Michael Jacobson, “Swim study reveals a smart pool of talent – Griffith News.” [Online]. Available: https://app.secure.griffith.edu.au/news/2013/08/13/swimming-a-smart-move-for-children/.
[4] S.Thuret, “You can grow new brain cells. Here’s how.,” TED Talk | TED.com, 2015. [Online]. Available: https://www.ted.com/talks/sandrine_thuret_you_can_grow_new_brain_cells_here_s_how/transcript?share=11ab6178f9.
[5] P.Oja, P.Kelly, Z.Pedisic, S.Titze, A.Bauman, C.Foster, M.Hamer, M.Hillsdon, andE.Stamatakis, “Associations of specific types of sports and exercise with all-cause and cardiovascular-disease mortality: a cohort study of 80 306 British adults,” Br. J. Sports Med., vol. 51, no. 10, pp. 812–817, 2017.
[6] 官筱芬, “基於三軸加速度計之游泳運動追蹤演算法開發,” 國立臺灣科技大學, 2017.
[7] B.Guignard, A.Rouard, D.Chollet, andL.Seifert, “Behavioral dynamics in swimming: The appropriate use of inertial measurement units,” Front. Psychol., vol. 8, no. MAR, pp. 1–16, 2017.
[8] R.Fernandes, J.Ribeiro, P.Figueiredo, L.Seifert, andJ.Vilas-Boas, “Kinematics of the hip and body center of mass in front crawl,” J. Hum. Kinet., vol. 33, no. 1, pp. 15–23, 2012.
[9] A. J.Callaway, “Measuring kinematic variables in front crawl swimming using accelerometers: A validation study,” Sensors (Switzerland), vol. 15, no. 5, pp. 11363–11386, 2015.
[10] R.Mooney, G.Corley, A.Godfrey, L.Quinlan, andG.ÓLaighin, “Inertial Sensor Technology for Elite Swimming Performance Analysis: A Systematic Review,” Sensors, vol. 16, no. 1, p. 18, 2015.
[11] Y.Ohgi, “Microcomputer-based Acceleration Sensor Device for Sports Biomechanics,” in Proceedings of IEEE Sensors Conference, 2002, pp. 699–704.
[12] N. P.Davey, M. E.Anderson, andD. A.James, “An accelerometer-based system for elite athlete swimming performance analysis,” Proc. SPIE, vol. 5649, p. 409, 2005.
[13] P.Siirtola, P.Laurinen, J.Röning, andH.Kinnunen, “Efficient Accelerometer-Based Swimming Exercise Tracking ,” IEEE Symp. Comput. Intell. Data Min., pp. 156–161, 2011.
[14] W.Choi, J.Oh, T.Park, S.Kang, M.Moon, U.Lee, I.Hwang, andJ.Song, “MobyDick,” Proc. 12th ACM Conf. Embed. Netw. Sens. Syst. - SenSys ’14, pp. 76–90, 2014.
[15] M. S.Pan, K. C.Huang, T. H.Lu, andZ. Y.Lin, “Using accelerometer for counting and identifying swimming strokes,” Pervasive Mob. Comput., vol. 31, pp. 37–49, 2016.
[16] U.Jensen, P.Blank, P.Kugler, andB. M.Eskofier, “Unobtrusive and Energy-Efficient Swimming Exercise Tracking Using On-Node Processing,” IEEE Sens. J., vol. 16, no. 10, pp. 3972–3980, 2016.
[17] K. C.Huang, C. P.Chu, T. K.Chiu, andJ. C.Chen, “Comparison of sensor analysis for swimming in different positions,” Proc. 2017 IEEE Int. Conf. Appl. Syst. Innov. Appl. Syst. Innov. Mod. Technol. ICASI 2017, pp. 1484–1487, 2017.
[18] R.Mooney, L. R.Quinlan, G.Corley, A.Godfrey, C.Osborough, andG.ÓLaighin, “Evaluation of the Finis Swimsense® and the Garmin SwimTM activity monitors for swimming performance and stroke kinematics analysis,” PLoS One, vol. 12, no. 2, 2017.
[19] S.Ganzevles, R.Vullings, P. J.Beek, H.Daanen, andM.Truijens, “Using tri-axial accelerometry in daily elite swim training practice,” Sensors (Switzerland), vol. 17, no. 5, 2017.
[20] “Gyroscope - Wikipedia.” [Online]. Available: https://en.wikipedia.org/wiki/Gyroscope.
[21] L.Breiman, “Random forests,” Mach. Learn., vol. 45, no. 1, pp. 5–32, 2001.
[22] C.~Cortes andV.~Vapnik, “Support Vector Networks,” Mach. Learn., vol. 20, no. 3, p. 273~-~297, 1995.
[23] “Apple Watch 38mm Aluminum Case Sport with Black Sport Band.” [Online]. Available: https://www.apple.com/shop/buy-watch/apple-watch/space-gray-aluminum-gray-sport-band?preSelect=false&product=MR352LL/A&step=detail#.
[24] “Swift - Apple (BR).” [Online]. Available: https://www.apple.com/swift/.
[25] Julius O. Smith III, “Direct-Form I.” [Online]. Available: https://ccrma.stanford.edu/~jos/fp/Direct_Form_I.html.
[26] B.Jensen, Ulf and Ring, Mathias and Eskofier, “Generic Features for Biosignal Classification,” Sportinformatik, pp. 162–168, 2012.
[27] I.Guyon andA.Elisseeff, “An Introduction to Variable and Feature Selection,” J. Mach. Learn. Res., vol. 3, no. 3, pp. 1157–1182, 2003.
[28] “scikit-learn: machine learning in Python — scikit-learn 0.19.1 documentation,” Scikit-Learn, 2017. [Online]. Available: http://scikit-learn.org/stable/.
[29] Y.Omae, Y.Kon, M.Kobayashi, K.Sakai, A.Shionoya, H.Takahashi, T.Akiduki, K.Nakai, N.Ezaki, Y.Sakurai, andC.Miyaji, “Swimming style classification based on ensemble learning and adaptive feature value by using inertial measurement unit,” J. Adv. Comput. Intell. Intell. Informatics, vol. 21, no. 4, 2017.
[30] C.-C.Chang andC.-J.Lin, “Libsvm,” ACM Trans. Intell. Syst. Technol., vol. 2, no. 3, pp. 1–27, 2011.
[31] “Forerunner 920XT | Garmin.” [Online]. Available: https://buy.garmin.com/en-US/US/p/137024.