隨著休閒運動風氣的盛行，無論是職業還是業餘愛好者，自行車是世界上相當受歡迎的運動項目。但是使用者往往都會忽略掉單車規格是否合適自身的尺寸，使自己在騎乘自行車時，處於不適當的關節角度運動，卻沒有發覺。長期而言，對身體的傷害會慢慢累積。因此，需根據單車適配的準則協助騎士找到適合自己身形的單車配置，透過動作捕捉系統來獲取關節角度，改善騎行姿勢來避免不必要的運動傷害。而影像辨識OpenPose可以藉由影像獲取人體骨架，成本低且方便性高，在運動姿勢的調整上也已有廣泛應用，能夠透過輸出關節點的位置分析運動動作的過程，改善不正確的運動姿勢。
現今單車適配需依靠動作捕捉系統和適配教練與適配者騎行的回饋進行動作分析，而本研究目標是利用單一攝影機的RGB影像進行單車適配，透過OpenPose的資訊預測三維人體骨架，其模型平均關節點距離(Mean Per Joint Position Error ,MPJPE)與平均空間角度誤差分別為76.34毫米和9.12度。並進一步假設人體肢段長度不變與腳尖踩踏位置固定，以髖關節中心與手腕關節作為調整的依據，模擬每調整一公分座墊高度或手把距離的骨架，使違反關節角度建議範圍的數量最小化。而調整後平均座墊高度提高1.1公分，手把距離則是向前0.1公分，調整後違反關節角度畫面幀數有統計顯著減少(p=0.016)。然而，經由騎行疲勞量表與視覺類比量表，得到受測者靜態方法調整與影像式優化調整的主觀感受，其調整前後疲勞與疼痛的數值皆無統計顯著差異(p>0.05)。
本研究建立一套單車適配的系統，利用影像獲取人體關節角度，加速單車適配所需的時間，給予適配者建議的調整方式，能夠減少不適當的關節角度，進而降低騎行時所可能累積的疲勞性運動傷害。而在其他運動亦能夠依照此系統的方法，根據不同運動的姿勢準則，給予使用者在運動動作上的建議。

With the popularity of recreational sports, cycling is one of the most popular sports in the world, whether professional or amateur. However, people often ignore whether the bicycle specifications are suitable for their size. Put user-self at an inappropriate joint angle while riding a bike without noticing. Over the long term, the damage to the body slowly accumulates. Therefore, it is necessary to assist the rider in finding a bicycle configuration that suits the body shape according to the principles of bike fitting. The motion capture system obtains the joint angle and improves the riding posture to avoid unnecessary sports injuries. The image recognition OpenPose can obtain the human skeleton through the image, which is low cost and high convenience. It has also been widely used in the adjustment of movement posture. It can analyze the movement process through the output of the joint position and improve the incorrect movement posture. Nowadays, bike fitting needs to rely on the motion capture system and the feedback from the bike fitting coach and the fitter to analyze the movements.
The study aimed is to focus on bike fitting using RGB image analysis from a single camera. The 3D human skeleton was predicted through the information of OpenPose. The mean per joint position error(MPJPE) and the average spatial angle error of the model was 76.34 mm and 9.12 degrees, respectively. Based on the assumption that the length of the human body remains unchanged and the toe position is fixed. Using the center of the hip joint and the wrist joint as the basis for adjustment, simulated a skeleton that adjusts the seat height or the handlebar distance by one centimeter. Minimized the number of joint angles that violated the recommended range, while the average seat height was increased by 1.1 cm after adjustment, and the average handlebar distance was 0.1 cm forward. There was a statistically significant reduction in the number of joint angles that violated the recommended range after adjustment (p=0.016). The subjects' subjective feelings of static method adjustment and image-based optimization adjustment were obtained through the OMNI Scale and the Visual Analogue Scale. However, there was no statistically significant difference in the numerical values of fatigue and pain between pre- and post-adjustment (p>0.05).
This study establishes a bike fitting system, which uses images to obtain the joint angle of the human body, and accelerate the time required for bike fitting. The recommended adjustment length for bike fitting can reduce the inappropriate joint angle, thereby reducing the overuse injuries accumulated when riding. In other sports, according to the method of this system, the user can be given suggestions on sports actions according to the posture criteria of different sports.

摘要 i
Abstract ii
目錄 iv
圖目錄 vi
表目錄 vii
第一章 緒論 1
1.1研究背景與動機 1
1.2研究目的與範圍 3
1.3研究架構與流程 3
第二章 文獻回顧 5
2.1單車傷害 5
2.1.1創傷性運動傷害 5
2.1.2疲勞性運動傷害 6
2.2單車適配 8
2.2.1靜態分析 9
2.3.2動態分析 9
2.3動作捕捉技術 16
2.3.1光學式 17
2.3.2影像式 19
2.4人體姿態估計在其他領域之應用 22
2.5單車主觀量表 27
2.5.1騎行疲勞量表 27
2.5.2視覺類比量表 29
2.6小結 31
第三章 研究方法 32
3.1研究對象 32
3.2實驗步驟 32
3.3研究工具 35
3.4資料處理 37
3.5系統架構 40
3.5.1騎乘姿勢3D模型 40
3.5.2 影像式單車適配優化 41
3.6統計分析 45
第四章 實驗結果 46
4.1受測者資訊 46
4.2模型結果 48
4.2.1關節點距離誤差 48
4.2.2關節角度 51
4.3主觀量表結果 55
第五章 討論 57
5.1騎行模型成果 57
5.2主觀量表成效 61
5.3單車適配建議應用 63
5.4研究限制 64
第六章 結論與未來方向 65
參考文獻 66
附錄一 研究倫理審查結果通知書 73

中文部分
1. 臺北市政府交通局(2020)，109年全年度交通事故分析報告。
取自：https://www.roadsafety.taipei/News_Content.aspx?n=55E08C53904C54
B4&sms=478BDD9A52935513&s=B4000DC51A937E73
2. 張特彰、賴仲亮、蔡明妙、呂傳欽、許碧珊(2009)腳踏車運動引起之傷害及預防。基層醫療；24(12)：439-443。
取自：https://www.tafm.org.tw/ehc-tafm/s/w/ebook/index_other/journalContent
1/663
3. optitrack動作捕捉系統-MEMSTEC 麥思科技有限公司。
取自：https://www.memstec.com.tw/product.php?cat=104
4. Vicon系統Plug In Gait Full-Body model。
取自：https://docs.vicon.com/display/Nexus212/Full+body+modeling+with+
Plug-in+Gait
5. Bryton踏頻感測器。
取自：https://www.brytonsport.com/#/sensor?id=cadence
6. Human3.6M數據集。
取自：http://vision.imar.ro/human3.6m/description.php

外文部分
1. Asplund, C., Webb, C., & Barkdull, T. (2005). Neck and back pain in bicycling. Current sports medicine reports, 4(5), 271-274.
2. Attewell, R. G., Glase, K., & McFadden, M. (2001). Bicycle helmet efficacy: a meta-analysis. Accident Analysis & Prevention, 33(3), 345-352.
3. Bailey, M., MAILLARDET, F., & Messenger, N. (2003). Kinematics of cycling in relation to anterior knee pain and patellar tendinitis. Journal of sports sciences, 21(8), 649-657.
4. Balan, A. O., Sigal, L., & Black, M. J. (2005). A quantitative evaluation of video-based 3D person tracking. Paper presented at the 2005 IEEE International Workshop on Visual Surveillance and Performance Evaluation of Tracking and Surveillance.
5. Bini, R., Hume, P. A., & Croft, J. L. (2011). Effects of bicycle saddle height on knee injury risk and cycling performance. Sports Medicine, 41(6), 463-476.
6. Bini, R. R., & Hume, P. (2016). A comparison of static and dynamic measures of lower limb joint angles in cycling: Application to bicycle fitting. Human Movement, 17(1), 36-42.
7. Blocken, B., Defraeye, T., Koninckx, E., Carmeliet, J., & Hespel, P. (2013). CFD simulations of the aerodynamic drag of two drafting cyclists. Computers & Fluids, 71, 435-445.
8. Bohlmann, J. T. (1981). Injuries in competitive cycling. The Physician and sportsmedicine, 9(5), 117-124.
9. Cao, Z., Hidalgo, G., Simon, T., Wei, S.-E., & Sheikh, Y. (2019). OpenPose: realtime multi-person 2D pose estimation using Part Affinity Fields. IEEE transactions on pattern analysis and machine intelligence, 43(1), 172-186.
10. Chen, W., Jiang, Z., Guo, H., & Ni, X. (2020). Fall detection based on key points of human-skeleton using openpose. Symmetry, 12(5), 744.
11. Chen, Y.-L., & Liu, Y.-N. (2014). Optimal protruding node length of bicycle seats determined using cycling postures and subjective ratings. Applied Ergonomics, 45(4), 1181-1186.
12. Choi, J. S., Seo, J. W., Lee, J. S., Kim, J., & Tack, G. R. (2018). Effects of dynamic bike fitting by lower limb alignment on pedaling performance. Journal of Mechanics in Medicine and Biology, 18(08), 1840039.
13. Ci, H., Wang, C., Ma, X., & Wang, Y. (2019). Optimizing network structure for 3d human pose estimation. Paper presented at the Proceedings of the IEEE/CVF international conference on computer vision.
14. Connick, M. J., & Li, F.-X. (2013). The impact of altered task mechanics on timing and duration of eccentric bi-articular muscle contractions during cycling. Journal of Electromyography and Kinesiology, 23(1), 223-229.
15. Dettori, N. J., & Norvell, D. C. (2006). Non-traumatic bicycle injuries. Sports Medicine, 36(1), 7-18.
16. Diefenthaeler, F., Bini, R. R., Karolczak, A. P. B., & Carpes, F. P. (2008). Muscle activation during pedaling in different saddle position. Revista Brasileira de Cineantropometria e Desempenho Humano, 10(2), 161-169.
17. Dsouza, G., Maurya, D., & Patel, A. (2020). Smart gym trainer using Human pose estimation. Paper presented at the 2020 IEEE International Conference for Innovation in Technology (INOCON).
18. Du, Y., Wong, Y., Liu, Y., Han, F., Gui, Y., Wang, Z., . . . Geng, W. (2016). Marker-less 3d human motion capture with monocular image sequence and height-maps. Paper presented at the European conference on computer vision.
19. Fintelman, D., Sterling, M., Hemida, H., & Li, F.-X. (2014). Optimal cycling time trial position models: Aerodynamics versus power output and metabolic energy. Journal of Biomechanics, 47(8), 1894-1898.
20. Gösele-Koppenburg, A. (2016). Cycling. In Foot and Ankle Sports Orthopaedics (pp. 441-445): Springer.
21. Grant, S., Aitchison, T., Henderson, E., Christie, J., Zare, S., Mc Murray, J., & Dargie, H. (1999). A comparison of the reproducibility and the sensitivity to change of visual analogue scales, Borg scales, and Likert scales in normal subjects during submaximal exercise. Chest, 116(5), 1208-1217.
22. Hamley, E., & Thomas, V. (1967). Physiological and postural factors in the calibration of the bicycle ergometer. The Journal of physiology, 191(2), 55P-56P.
23. He Ling, W., & Yun-Ju, L. (2019). Occupational evaluation with Rapid Entire Body Assessment (REBA) via imaging processing in field. Paper presented at the Human Factors Society Conference.
24. Heil, D., Derrick, T., & Whittlesey, S. (1997). The relationship between preferred and optimal positioning during submaximal cycle ergometry. European journal of applied physiology and occupational physiology, 75(2), 160-165.
25. Hinz, T., Navarro-Guerrero, N., Magg, S., & Wermter, S. (2018). Speeding up the hyperparameter optimization of deep convolutional neural networks. International Journal of Computational Intelligence and Applications, 17(02), 1850008.
26. Holliday, W., Fisher, J., Theo, R., & Swart, J. (2017). Static versus dynamic kinematics in cyclists: A comparison of goniometer, inclinometer and 3D motion capture. European journal of sport science, 17(9), 1129-1142.
27. Hwang, J., Park, S., & Kwak, N. (2017). Athlete pose estimation by a global-local network. Paper presented at the Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops.
28. Insafutdinov, E., Andriluka, M., Pishchulin, L., Tang, S., Levinkov, E., Andres, B., & Schiele, B. (2017). Arttrack: Articulated multi-person tracking in the wild. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
29. Insafutdinov, E., Pishchulin, L., Andres, B., Andriluka, M., & Schiele, B. (2016). Deepercut: A deeper, stronger, and faster multi-person pose estimation model. Paper presented at the European conference on computer vision.
30. Ionescu, C., Papava, D., Olaru, V., & Sminchisescu, C. (2013). Human3. 6m: Large scale datasets and predictive methods for 3d human sensing in natural environments. IEEE transactions on pattern analysis and machine intelligence, 36(7), 1325-1339.
31. Iqbal, U., & Gall, J. (2016). Multi-person pose estimation with local joint-to-person associations. Paper presented at the European conference on computer vision.
32. Jafarzadeh, P., Virjonen, P., Nevalainen, P., Farahnakian, F., & Heikkonen, J. (2021). Pose Estimation of Hurdles Athletes using OpenPose. Paper presented at the 2021 International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME).
33. Kamel, A., Liu, B., Li, P., & Sheng, B. (2019). An investigation of 3D human pose estimation for learning Tai Chi: A human factor perspective. International Journal of Human–Computer Interaction, 35(4-5), 427-439.
34. Kang, J., Hoffman, J. R., Walker, H., Chaloupka, E. C., & Utter, A. C. (2003). Regulating intensity using perceived exertion during extended exercise periods. European Journal of Applied Physiology, 89(5), 475-482.
35. Kotler, D. H., Babu, A. N., & Robidoux, G. (2016). Prevention, evaluation, and rehabilitation of cycling-related injury. Current sports medicine reports, 15(3), 199-206.
36. LeMond, G. (1987). Gordis K. Greg LeMond's Complete Book of Bicycling. In: New York, NY, Perigee Books.
37. Liparoti, M., Lopez, E. T., & Agosti, V. (2020). Motion capture system: A useful tool to study cyclist’s posture. Journal of Physical Education and Sport, 20(4), 2364-2367.
38. Mündermann, A., Nigg, B. M., Stefanyshyn, D. J., & Humble, R. N. (2002). Development of a reliable method to assess footwear comfort during running. Gait & posture, 16(1), 38-45.
39. Ma, X., Su, J., Wang, C., Ci, H., & Wang, Y. (2021). Context modeling in 3d human pose estimation: A unified perspective. Paper presented at the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.
40. Martinez, J., Hossain, R., Romero, J., & Little, J. J. (2017). A simple yet effective baseline for 3d human pose estimation. Paper presented at the Proceedings of the IEEE International Conference on Computer Vision.
41. Mehdizadeh, S., Arshi, A. R., & Davids, K. (2014). Effect of speed on local dynamic stability of locomotion under different task constraints in running. European journal of sport science, 14(8), 791-798.
42. Mehra, R. (2018). Breast cancer histology images classification: Training from scratch or transfer learning? ICT Express, 4(4), 247-254.
43. Mellion, M. B. (1991). Common cycling injuries. Sports Medicine, 11(1), 52-70.
44. Nair, V., & Hinton, G. E. (2010). Rectified linear units improve restricted boltzmann machines. Paper presented at the Icml.
45. Neptune, R. R., & Hull, M. L. (1996). Methods for determining hip movement in seated cycling and their effect on kinematics and kinetics. Journal of Applied Biomechanics, 12(4), 493-507.
46. Newell, A., Yang, K., & Deng, J. (2016). Stacked hourglass networks for human pose estimation. Paper presented at the European conference on computer vision.
47. Ota, M., Tateuchi, H., Hashiguchi, T., Kato, T., Ogino, Y., Yamagata, M., & Ichihashi, N. (2020). Verification of reliability and validity of motion analysis systems during bilateral squat using human pose tracking algorithm. Gait & posture, 80, 62-67.
48. Park, S., Hwang, J., & Kwak, N. (2016). 3d human pose estimation using convolutional neural networks with 2d pose information. Paper presented at the European Conference on Computer Vision.
49. Pavlakos, G., Zhou, X., Derpanis, K. G., & Daniilidis, K. (2017). Coarse-to-fine volumetric prediction for single-image 3D human pose. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
50. Peveler, W., Bishop, P., Smith, J., Richardson, M., & Whitehorn, E. (2005). COMPARING METHODS FOR SETTING SADDLE HEIGHT IN TRAINED CYCLISTS. Journal of Exercise Physiology Online, 8(1).
51. Pishchulin, L., Insafutdinov, E., Tang, S., Andres, B., Andriluka, M., Gehler, P. V., & Schiele, B. (2016). Deepcut: Joint subset partition and labeling for multi person pose estimation. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
52. Price, D., & Donne, B. (1997). Effect of variation in seat tube angle at different seat heights on submaximal cycling performance in man. Journal of sports sciences, 15(4), 395-402.
53. Priego Quesada, J. I., Kerr, Z. Y., Bertucci, W. M., & Carpes, F. P. (2019). A retrospective international study on factors associated with injury, discomfort and pain perception among cyclists. PloS one, 14(1), e0211197.
54. Priego Quesada, J. I., Pérez-Soriano, P., Lucas-Cuevas, A. G., Salvador Palmer, R., & Cibrián Ortiz de Anda, R. M. (2017). Effect of bike-fit in the perception of comfort, fatigue and pain. Journal of sports sciences, 35(14), 1459-1465.
55. Qiao, S., Wang, Y., & Li, J. (2017). Real-time human gesture grading based on OpenPose. Paper presented at the 2017 10th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI).
56. Rahul, M. (2018). Review on motion capture technology. Global Journal of Computer Science and Technology.
57. Rhea, M. R. (2004). Determining the magnitude of treatment effects in strength training research through the use of the effect size. The Journal of Strength & Conditioning Research, 18(4), 918-920.
58. Ribeiro, T. H., & Vieira, M. L. H. (2016). Motion Capture Technology—Benefits and Challenges. Int. J. Innov. Res. Technol. Sci. Int. J. Innov. Res. Technol. Sci, 48(1), 2321-1156.
59. Salai, M., Brosh, T., Blankstein, A., Oran, A., & Chechik, A. (1999). Effect of changing the saddle angle on the incidence of low back pain in recreational bicyclists. British journal of sports medicine, 33(6), 398-400.
60. Sanderson, D. J., & Amoroso, A. T. (2009). The influence of seat height on the mechanical function of the triceps surae muscles during steady-rate cycling. Journal of Electromyography and Kinesiology, 19(6), e465-e471.
61. Scoz, R. D., Amorim, C. F., Espindola, T., Santiago, M., Mendes, J. J. B., de Oliveira, P. R., . . . Brito, R. N. (2021). Discomfort, pain and fatigue levels of 160 cyclists after a kinematic bike-fitting method: an experimental study. BMJ open sport & exercise medicine, 7(3), e001096.
62. Silberman, M. R. (2013). Bicycling injuries. Current sports medicine reports, 12(5), 337-345.
63. Silberman, M. R., Webner, D., Collina, S., & Shiple, B. J. (2005). Road bicycle fit. Clinical Journal of Sport Medicine, 15(4), 271-276.
64. Streisfeld, G. M., Bartoszek, C., Creran, E., Inge, B., McShane, M. D., & Johnston, T. (2017). Relationship between body positioning, muscle activity, and spinal kinematics in cyclists with and without low back pain: A systematic review. Sports health, 9(1), 75-79.
65. Tao, H., Al-Sulttani, A. O., Salih Ameen, A. M., Ali, Z. H., Al-Ansari, N., Salih, S. Q., & Mostafa, R. R. (2020). Training and testing data division influence on hybrid machine learning model process: application of river flow forecasting. Complexity, 2020.
66. Thompson, M. J., & Rivara, F. P. (2001). Bicycle-related injuries. American Family Physician, 63(10), 2007.
67. Umberger, B. R., & Martin, P. E. (2001). Testing the planar assumption during ergometer cycling. Journal of Applied Biomechanics, 17(1), 55-62.
68. Van Hoof, W., Volkaerts, K., O'Sullivan, K., Verschueren, S., & Dankaerts, W. (2012). Comparing lower lumbar kinematics in cyclists with low back pain (flexion pattern) versus asymptomatic controls–field study using a wireless posture monitoring system. Manual therapy, 17(4), 312-317.
69. Viswakumar, A., Rajagopalan, V., Ray, T., & Parimi, C. (2019). Human gait analysis using OpenPose. Paper presented at the 2019 Fifth International Conference on Image Information Processing (ICIIP).
70. Wang, J., Huang, S., Wang, X., & Tao, D. (2019). Not all parts are created equal: 3d pose estimation by modeling bi-directional dependencies of body parts. Paper presented at the Proceedings of the IEEE/CVF international conference on computer vision.
71. Wei, S.-E., Ramakrishna, V., Kanade, T., & Sheikh, Y. (2016). Convolutional pose machines. Paper presented at the Proceedings of the IEEE conference on Computer Vision and Pattern Recognition.
72. Williamson, A., & Hoggart, B. (2005). Pain: a review of three commonly used pain rating scales. Journal of clinical nursing, 14(7), 798-804.
73. Xiaohan Nie, B., Wei, P., & Zhu, S.-C. (2017). Monocular 3d human pose estimation by predicting depth on joints. Paper presented at the Proceedings of the IEEE International Conference on Computer Vision.
74. Yang, W., Ouyang, W., Wang, X., Ren, J., Li, H., & Wang, X. (2018). 3d human pose estimation in the wild by adversarial learning. Paper presented at the Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.
75. Zecha, D., Einfalt, M., Eggert, C., & Lienhart, R. (2018). Kinematic pose rectification for performance analysis and retrieval in sports. Paper presented at the Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops.
76. Zhang, Z., Li, S., Wan, B., Visentin, P., Jiang, Q., Dyck, M., . . . Shan, G. (2016). The influence of X-factor (trunk rotation) and experience on the quality of the badminton forehand smash. Journal of human kinetics, 53, 9.
77. Zhao, L., Peng, X., Tian, Y., Kapadia, M., & Metaxas, D. N. (2019). Semantic graph convolutional networks for 3d human pose regression. Paper presented at the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.
78. Zhou, X., Huang, Q., Sun, X., Xue, X., & Wei, Y. (2017). Towards 3d human pose estimation in the wild: a weakly-supervised approach. Paper presented at the Proceedings of the IEEE International Conference on Computer Vision.
79. Zhou, X., Sun, X., Zhang, W., Liang, S., & Wei, Y. (2016). Deep kinematic pose regression. Paper presented at the European Conference on Computer Vision.