本研究使用計算機視覺及視覺人工智慧針對目標進行偵測、追蹤與運動分析，並應用於運動學的動作分析和鑄造業的澆鑄操作。針對前者，應用OpenPose和模糊動作評估器開發一項無標記的運動評估系統。無標記技術在定位關節位置及骨架形成方面具有較佳的準確性，模糊動作評估系統用來協助處理肘關節外展或內收可能引起的定位偏差。該系統在模糊推理的支持下得到了定量和定性的評估，進行而產生令人滿意和可靠的分析結果。
在澆注作業監控方面，主要目的為評估作業期間，盛鋼桶澆注角度的變化。由於鑄造過程中，熔融金屬產生的高溫、高熱及高亮度，增加了視覺分析的難度，本研究採用計算機視覺與視覺人工智慧的方法挑戰影像過曝的問題。使用計算機視覺技術，經由目標偵測、追蹤、去背、Canny邊緣檢測、霍夫直線偵測等影像處理，進行角度估算。在視覺人工智慧方面，運用方向梯度直方圖(HOG)結合支持向量機對模型進行訓練，偵測與識別每一幀的盛鋼桶，接著對識別的盛鋼桶進行標注點的位置預測，從而得到估算的角度。
總體而言，應用無標計監控評估器的技術，得到動作上高度準確的分析結果，相比之下，在高亮度所引起影像過度曝光的鑄造環境下，對澆注作業更具挑戰性的及時監測技術，則取得了可靠的角度預測成果。

This study used computer vision and vision AI for object detection, tracking, and motion analysis with applications in sports kinematics and casting operations in manufacturing. Regarding the former application, developing a marker-free movement assessment system through OpenPose and fuzzy action assessor. The marker-free technique produces excellent accuracy in localizing joint body positions and forming the skeleton. A fuzzy action assessor is used to assist with handling instability that is probably by abduction or adduction at the elbow conjunct. The evaluation system acquired quantitative and qualitative assessments considered by the supports of fuzzy inference and generated satisfactory and dependable results.
Concerning monitoring the casting operation, the main objective was to estimate the pouring angles of a dipper during the pouring work. Due to the high temperature and over-exposed environment during casting, this study employed computer vision and vision AI to over the measuring difficulty. When using computer vision, the processing of object detection, tracking, foreground extraction, Canny edge detection, and Hough transform will perform.
In terms of using vision AI, the histogram of oriented gradient (HOG) combines with the support vector machine (SVM) for data classification and use to object extraction and recognition. More specifically, landmarks tracked for the pouring ladle were closely examined and analyzed.
Overall, the techniques applied to a marker-free monition evaluator achieved great success. In comparison, employing the online monitoring of the pouring operation under a high-temperature and over-exposed casting environment was much more challenging, even though some degree of success had met achieved.

Table of Contents
中文摘要 -----------------------------------------------------------------------ⅰ
Abstract ----------------------------------------------------------------------ⅱ
Acknowledgement --------------------------------------------------------------ⅲ
Contents ---------------------------------------------------------------------ⅳ
List of tables ---------------------------------------------------------------ⅵ
List of Figures -------------------------------------------------------------ⅷ
1、Introduction ---------------------------------------------------------------1
1.1 Human motion analysis ---------------------------------------------------2
1.2 Object motion analysis --------------------------------------------------5
2、Literature review ----------------------------------------------------------9
2.1 Human motion analysis ---------------------------------------------------9
2.1.1 Convolutional Neural Network (CNN) structure: Visual Geometry Group 19
(VGG19) -------------------------------------------------------------------9
2.1.2 Fuzzy inference ----------------------------------------------------10
2.1.3 OpenPose -----------------------------------------------------------11
2.2 Object motion analysis -------------------------------------------------14
2.2.1 Computer vision ----------------------------------------------------14
2.2.1.1 GrabCut --------------------------------------------------------14
2.2.1.2 Tracking-Learning Detection ------------------------------------17
2.2.1.3 Canny Edge Detection -------------------------------------------25
2.2.1.4 Lines detection with Hough transform ---------------------------27
2.2.2 Compute AI ---------------------------------------------------------29
2.2.2.1 Histogram of Oriented Gradients --------------------------------29
2.2.2.2 Support Vector Machine -----------------------------------------29
2.2.2.3 Ensemble of Regression Trees -----------------------------------33
3、Research Method and Research Design ---------------------------------------35
3.1 Push-up motion evaluation ----------------------------------------------36
3.1.1 Participants -------------------------------------------------------36
3.1.2 Design and procedure -----------------------------------------------36
3.2 A ladle pouring motion evaluation --------------------------------------38
3.2.1 Design and procedure -----------------------------------------------38
4、Data analysis -------------------------------------------------------------52
4.1 Push-up motion ---------------------------------------------------------52
4.1.1 Data collection ----------------------------------------------------52
4.1.2 Data processing ----------------------------------------------------52
4.1.3 Data metrics -----------------------------------------------------------58
4.1.4 Results ----------------------------------------------------------------58
4.1.5 Discussion -------------------------------------------------------------62
4.2 A ladle pouring motion -------------------------------------------------65
4.2.1 Data collection ----------------------------------------------------65
4.2.2 Data processing-----------------------------------------------------66
4.2.3 Data metrics -------------------------------------------------------69
4.2.4 Results ------------------------------------------------------------69
4.2.5 Discussion ---------------------------------------------------------80
5、Conclusion ----------------------------------------------------------------91
5.1 About push-up motion evaluation ----------------------------------------91
5.2 A ladle pouring motion -------------------------------------------------92
Reference --------------------------------------------------------------------93

List of tables
Table 1. The dataset of human motion analysis -------------------------------52
Table 2. The input and output for UP movement -------------------------------57
Table 3. The input and output for DOWN movement -----------------------------57
Table 4. The comparison of the presented markerfree approaches with others --64
Table 5. The dataset of a ladle pouring motion analysis –--------------------66
Table 6. Training result of the model of 2 points and 5 points --------------70
Table 7. Training result of two types of models -----------------------------71
Table 8. The results of two types of models' object recognition -------------71
Table 9. The two types of models' the accuracy of the calculated angle ------71
Table 10. Calculated angles' results ----------------------------------------72
Table 11. Compare three types of models' results ----------------------------72
Table 12. Description analysis of three methods -----------------------------73
Table 13. The result of ANOVA -----------------------------------------------73
Table 14. TUKEY HSD/KRAMER Tests --------------------------------------------74
Table 15. Q TEST ------------------------------------------------------------74
Table 16. Levene's Tests ----------------------------------------------------75
Table 17. Training result of the model for shuffling and sorting ------------75
Table 18. Training result of two models -------------------------------------76
Table 19. Ladle recognition -------------------------------------------------76
Table 20. The accuracy of the calculated angles -----------------------------76
Table 21. T-test for shuffling vs. Sorting ----------------------------------77
Table 22. The accuracy of the different models ------------------------------77
Table 23. T-test for 2 points vs. Shuffling ---------------------------------78
Table 24. T-test for 5 points(bottom) vs. Shuffling -------------------------78
Table 25. T-test for 5 points(ear) vs. Shuffling ----------------------------78
Table 26. T-test for 2 points(ear) vs. Sorting ------------------------------79
Table 27. T-test for 5 points(bottom) vs. Sorting ---------------------------79
Table 28. T-test for 5 points(ear) vs. Sorting ------------------------------79
Table 29. The difference between analysis of computer vision and machine
learning. ---------------------------------------------------------90

List of figures
Figure 1. Motion analysis architecture diagram -------------------------------1
Figure 2. The motion analysis for marker-based and markerless-based ----------4
Figure 3. The processes of casting -------------------------------------------5
Figure 4. The defective products ---------------------------------------------7
Figure 5. VGG 16 architecture ------------------------------------------------9
Figure 6. Fuzzy inference architecture --------------------------------------10
Figure 7. The refined PAF ---------------------------------------------------12
Figure 8. The prediction of confidence maps ---------------------------------13
Figure 9. CMU panoptic studio -----------------------------------------------14
Figure 10. An interactive function of GrabCut -------------------------------15
Figure 11. The results of the border matting algorithm ----------------------16
Figure 12. The moving object’s appearances are hugely different over time----18
Figure 13. A tracker of TLD -------------------------------------------------20
Figure 14. A detector of TLD ------------------------------------------------21
Figure 15. The learning of P-N experts --------------------------------------22
Figure 16. Learning of TLD --------------------------------------------------23
Figure 17. The TLD architecture ---------------------------------------------24
Figure 18. The procedure of Canny edge detection ----------------------------25
Figure 19. Hysteresis Thresholding discriminates the edges ------------------27
Figure 20. Hough transform for detecting a line -----------------------------28
Figure 21. The mapping grids of the Hough transform -------------------------29
Figure 22. A HOG representation ---------------------------------------------30
Figure 23. The optimal nonlinear decision boundary of SVM outputting --------31
Figure 24. (a) The C value is too small to cause under-fitting, (b) The optimal
C value obtains appropriate fitting, (c) The C value is too large to
cause over-fitting -----------------------------------------------32
Figure 25. (a) A flat 2D surface distributed data, (b) A hyperplane distributed
data -------------------------------------------------------------33
Figure 26. The next regression function -------------------------------------34
Figure 27. The research architecture ----------------------------------------35
Figure 28. A camera placement for push-up -----------------------------------37
Figure 29. A camera placement for pouring operation -------------------------38
Figure 30. The procedures of computer vision of an object -------------------39
Figure 31. (a) A pouring ladle operated by an operator standing, (b) The ladle
bounded by the selected ROI --------------------------------------40
Figure 32. Foreground extraction --------------------------------------------41
Figure 33. Binary processing ------------------------------------------------41
Figure 34. Canny edge detection ---------------------------------------------42
Figure 35. Hough lines transform --------------------------------------------42
Figure 36. The pitch angle of a ladle ---------------------------------------43
Figure 37. A cut ladle ------------------------------------------------------45
Figure 38. (a) Sobel X Gradients, (b) Sobel Y Gradients ---------------------45
Figure 39. Magnitudes and orientations of gradients -------------------------46
Figure 40. A cell of HOG ----------------------------------------------------46
Figure 41. The ballot box of HOG --------------------------------------------47
Figure 42. The feature vectors of HOG of the transformed ladle --------------47
Figure 43. The initial shape ------------------------------------------------48
Figure 44. An ensemble of regression trees ----------------------------------49
Figure 45. Three sampling methods -------------------------------------------50
Figure 46. (a) and (b) The number of ladles' landmarks of two and five,
respectively. ----------------------------------------------------50
Figure 47. (a) and (b) The positions of ladles' landmarks of ear and bottom,
respectively. ----------------------------------------------------51
Figure 48. (a) Synoptic graph for the locations of keypoints (b) Synoptic graph
for the body skeletons (c) Synoptic graph for UP movement's three
joint angles (d) Synoptic graph for DOWN movement's three joint
angles -----------------------------------------------------------53
Figure 49. (a) Membership Functions for Input Variable (Angle1) (b) Membership
Functions for Input Variable (Angle2) (c) Membership Functions for
Input Variable (Angle3) (d) Membership Functions for Output Variable
(Down pose) ------------------------------------------------------57
Figure 50. (a) The confusion matrix of UP poses: Coaches' determination vs.
Openpose with Fuzzy assessor (b) The confusion matrix of DOWN poses:
Coaches' determination vs. Openpose with Fuzzy assessor ----------59
Figure 51. (a) The UP poses of using the fuzzy action assessor (b) The DOWN
poses of using the fuzzy action assessor -------------------------60
Figure 52. (a) The chart of a qualified UP movement by OpenPose rendering, (b)
The chart of a qualified DOWN movement by OpenPose rendering -----61
Figure 53. (a) The chart of an unqualified UP movement by OpenPose rendering;
(b) The chart of an unqualified DOWN movement by OpenPose rendering
------------------------------------------------------------------63
Figure 54. Scheme diagram for the data collection of the pouring operation --65
Figure 55. The optimal line on a ladle of a frame ---------------------------67
Figure 56. The pitch angle of a ladle ---------------------------------------67
Figure 57. The landmarks on a ladle -----------------------------------------68
Figure 58. The pitch angle of a ladle ---------------------------------------68
Figure 59. The pitch angle of a ladle ---------------------------------------70
Figure 60. The comparison results of 2 points between 5 points on a ladle ---80
Figure 61. The comparison results of the ear and the bottom at the ladle's
positions --------------------------------------------------------81
Figure 62. The comparison results of the different sampling arrangements ----82
Figure 63. The ANOVA results of the different sampling arrangements ---------83
Figure 64. The Post-hoc results of the different sampling arrangements ------84
Figure 65. The comparison results of the shuffling and the sorting on the
samplings --------------------------------------------------------85
Figure 66. The comparison results of the different sampling arrangements ----86
Figure 67. The T-test results of the shuffling and sorting ------------------87
Figure 68. The T-test results of the shuffling and different sampling
arrangements -----------------------------------------------------88
Figure 69. The T-test results of the sorting and different sampling
arrangements -----------------------------------------------------89

[1] Lin, H.‐T., F.-C. Su, M. Nakamura, and Y. S. Chao, 2003, “Complex Chain of Momentum Transfer of Body Segments in the Baseball Pitching Motion”, Journal of the Chinese Institute of Engineers, vol. 26(6), pp.861-868, March. doi.org/10.1080/02533839.2003.9670841
[2] Bahadori, S., P. Davenport, T. Immins, and T. W. Wainwright, 2019, “Validation of Joint Angle Measurements: Comparison of a Novel Low-Cost Marker-Less System with an Industry Standard Marker-Based”, Journal of Medical Engineering and Technology, vol.43(1), pp.19-24, April. doi.org/10.1080/03091902.2019.1599072
[3] Zhou, H., and H. Hu, 2008,” Human Motion Tracking for Rehabilitation – A Survey”, Biomedical Signal Processing and Control, vol.3(1), pp.1–18, January. doi.org/10.1016/j.bspc.2007.09.001
[4] Fern'ndez-Baena, A., A. Susín, and X. Lligadas, 2012, “Biomechanical Validation of Upper-Body and Lower-Body Joint Movements of Kinect Motion Capture Data for Rehabilitation Treatments”, In 2012 Fourth International Conference on Intelligent Networking and Collaborative Systems, Bucharest, Romania, pp. 656-661, 19-21 September 2012. doi.org/10.1109/iNCoS.2012.66
[5] Pfister, A., A. M. West, S. Bronner, and J. A. Noah, 2014, “Comparative Abilities of Microsoft Kinect and Vision 3D Motion Capture for Gait Analysis”, Journal of Medical Engineering and Technology, vol.38(5), pp.274-280, May. doi.org/10.3109/03091902.2014.909540
[6] Cappozzo, A., F. Catani, A. Leardini, M. G. Benedetti, and C. U. Della, 1996, ”Position and Orientation in Space of Bones During Movement: Experimental Artefacts”, Clinical Biomechanics, vol.11(2), pp.90-100, March. doi.org/10.1016/0268-0033(95)00046-1
[7] Fuller, J., L.-J. Liu, M. C. Murphy, and R. W. Mann, 1997, “A Comparison of Lower-Extremity Skeletal Kinematics Measured Using Skin- and Pin-mounted Markers”, Human Movement Science, vol.16(2-3), pp.219-242, April. doi.org/10.1016/S0167-9457(96)00053-X
[8] Corazza, S., L. Mündermann, A. M. Chaudhari, T. Demattio, C. Cobelli, and T. P. Andriacchi, 2006, “A Markerless Motion Capture System to Study Musculoskeletal Biomechanics: Visual Hull and Simulated Annealing Approach”, Annals Biomedical Engineering, vol.34(6), pp.1019-1029, May. doi.org/10.1007/s10439-006-9122-8
[9] Croce, U.D., A. Leardini, L. Chiari, and A. Cappozzo, 2005, “Human Movement Analysis Using Stereophotogrammetry. Part 4: Assessment of Anatomical Landmark Misplacement and its Effects on Joint Kinematics”, Gait and Posture, vol.21(2), pp.226–237, February. doi.org/10.1016/j.gaitpost.2004.05.003
[10] Chiari, L., C. U. Della, A. Leardini, and A. Cappozzo, 2005, “Human Movement Analysis Using Stereophotogrammetry. Part 2: Instrumental Errors”, Gait Posture, vol.21(2), pp.197-211, February. doi.org/10.1016/j.gaitpost.2004.04.004
[11] Taylor, W.R., R. M. Ehrig, G. N. Duda, H. Schell, P. Seebeck, and M. O. Heller, 2006, “On the Influence of Soft Tissue Coverage in the Determination of Bone Kinematics Using Skin Markers”, Journal of Orthopaedic Research, vol.23(4), pp.726-734, January. doi.org/10.1016/j.orthres.2005.02.006
[12] Lu, T.W., and J. J. O'Connor, 1999, “Bone Position Estimation from Skin Marker Co-ordinates Using Global Optimisation with Joint Constraints”, Journal of Biomechanics, vol.32(2), pp.129-134, February. doi.org/10.1016/s0021-9290(98)00158-4
[13] Stagni, R., S. Fantozzi, and A. Cappello, 2009, “Double Calibration vs. Global Optimisation: Performance and Effectiveness for Clinical Application”, Gait Posture, vol.29(1), pp.119-122, January. doi.org/10.1016/j.gaitpost.2008.07.008
[14] Duprey, S., L. Cheze, and R. Dumas, 2010, “Influence of Joint Constraints on Lower Limb Kinematics Estimation from Skin Markers Using Global Optimization”, Journal of Biomechanics, vol.43(14), pp.2858-2862, October. doi.org/10.1016/j.jbiomech.2010.06.010
[15] Andersen, M.S., D. L. Benoit, M. Damsgaard, D. K. Ramsey, and J. Rasmussen, 2010, “Do Kinematic Models Reduce the Effects of Soft Tissue Artefacts in Skin Marker-Based Motion Analysis? An in Vivo Study of Knee kinematics”, Journal of Biomechanics, vol.43(2), pp.268-273, January. doi.org/10.1016/j.jbiomech.2009.08.034
[16] Mündermann, L., S. Corazza, and T. P. Andriacchi, 2006, “The Evolution of Methods for the Capture of Human Movement Leading to Markerless Motion Capture for Biomechanical Applications”, Journal of Neuroengineering and Rehabilitation, vol.3, pp.6, March. doi.org/10.1186/1743-0003-3-6
[17] Moeslund, T.B., and E. Granum, 2001, “A Survey of Computer Vision-Based Human Motion Capture”, Computer Vision and Image Understanding, vol.81(3), pp.231-268, March. doi.org/10.1006/cviu.2000.0897
[18] Clark, R.A., Y.-H. Pua, A. L. Bryant, and M. A. Hunt, 2013, “Validity of the Microsoft Kinect for Providing Lateral Trunk Lean Feedback During Gait Retraining”, Gait Posture, vol.38(4), pp.1064-1066, September. doi.org/10.1016/j.gaitpost.2013.03.029
[19] Chang, Y.-J., S.-F. Chen, and J.-D. Huang, 2011, “A Kinect-Based System for Physical Rehabilitation: A Pilot Study for Young Adults with Motor Disabilities”, Research in Developmental Disabilities, vol.32(6), pp.2566-2570, November-December. doi.org/10.1016/j.ridd.2011.07.002
[20] Raptis, M., D. Kirovski, and H. Hoppe, 2001, “Real-Time Classification of Dance Gestures from Skeleton Animation”, In SCA'11: Proceedings of the 2011 ACM SIGGRAPH/ Eurographics Symposium on Computer Animation, pp. 147–156, Vancouver, British Columbia, Canada, 5 – 7 August 2011. doi.org/10.1145/2019406.2019426
[21] Nambiar, A.M., P. L. Correia, and L. D. Soares, 2012, “Frontal Gait Recognition Combining 2D and 3D Data”, In MM&Sec '12: Proceedings of Multimedia and Security, pp. 145–150, Coventry, United Kingdom, 6 – 7 September 2012. doi.org/10.1145/2361407.2361432
[22] Stone, E., and M. Skubic, 2011, “Evaluation of an Inexpensive Depth Camera for In-Home Gait Assessment”, Journal of Ambient Intelligence and Smart Environments, vol.3(4), pp.349–361, January. doi.org/10.3233/AIS-2011-0124
[23] Exell, T., C. Freeman, and K. Meadmore, 2013, “Goal Orientated Stroke Rehabilitation Utilising Electrical Stimulation, Iterative Learning and Microsoft Kinect”, In 2013 IEEE 13th International Conference on Rehabilitation Robotics (ICORR), pp. 1-6, Seattle, WA, USA, 24-26 June 2013. doi.org/10.1109/ICORR.2013.6650493
[24] Gama, A. D., T. Chaves, L. Figueiredo, and V. Teichrieb, 2012, “Guidance and Movement Correction Based on Therapeutics Movements for Motor Rehabilitation Support Systems”, In 2012 14th Symposium on Virtual and Augmented Reality, pp. 191-200, Rio de Janeiro, Brazil, 28-31 May 2012. doi.org/10.1109/SVR.2012.15
[25] Pastor, I., H. A. Hayes, and S. J. Bamberg, 2012, “A Feasibility Study of an Upper Limb Rehabilitation System Using Kinect and Computer Games”, In Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp.1286-1289, San Diego, CA, USA, 28 August - 01 September 2012. doi.org/10.1109/EMBC.2012.6346173
[26] Lee, G. 2013. “Effects of Training Using Video Games on the Muscle Strength, Muscle Tone, and Activities of Daily Living of Chronic Stroke Patients”, Journal of Physical Therapy Science, vol.25(5), pp.595-597, May. doi.org/10.1589/jpts.25.595
[27] Sin, H., and G. Lee, 2013, “Additional Virtual Reality Training Using Xbox Kinect in Stroke Survivors with Hemiplegia”, American Journal of Physical and Medicine Rehabilitation, vol.92(10), pp.871-880, October. doi.org/10.1097/PHM.0b013e3182a38e40
[28] Cao, Z., G. Hidalgo, T. Simon, S.-E. Wei, and Y. Sheikh, 2021, “OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields”, IEEE Transactions on Pattern Analysis & Machine Intelligence, vol.43(1), pp.172-186, January. doi.org/10.1109/TPAMI.2019.2929257
[29] Chen, T.-T., and T.-C. Su, 2017, “Fuzzy-Based Decision-Making Applied to Performance Evaluation in Value Engineering”, Journal of the Chinese Institute of Engineers, vol.40(3), pp.200-206, March. doi.org/10.1080/02533839.2017.1300071
[30] Hasan A. and Suyitno, 2015, “Effect Pouring Temperature on Casting Defect Susceptibility of Hot Tearing in Metal Alloy Al-Si”, Applied Mechanics and Materials, vol.758, pp.95–99, April. doi.org/10.4028/www.scientific.net/amm.758.95
[31] Li, M., Y. Li, and H. Zhou, 2020, “Effects of Pouring Temperature on Microstructure and Mechanical Properties of the A356 Aluminum Alloy Diecastings”, Materials Research, vol.23 (1), pp.1-11, March. doi.org/10.1590/1980-5373-MR-2019-0676
[32] Ding, M., J. Song, and L. Honghui, 2014, “Effect of Pouring Temperature on Typical Structure of Thin-Walled ZL105A Alloy Casting”, Materials and Manufacturing Processes, vol.29(7), pp.853-863, July. doi.org/10.1080/10426914.2014.880472
[33] Mikell P. Groover, 2010, FUNDAMENTALS OF MODERNMANUFACTURING- Materials, Processes, and Systems, 4th Edition. John Wiley & Sons, Inc., New Jersey.
[34] Wisam M. Abu Jadayil, 2011, “Studying the Effects of Varying the Pouring Rate on the Casting Defects Using Nondestructive Testing Techniques”, Jordan Journal of Mechanical and Industrial Engineering, vol.5(6), pp.521-526, Number.
[35] Mohammad B. NDALIMAN and Akpan P. PIUS, 2007, “Behavior of Aluminum Alloy Castings under Different Pouring Temperatures and Speeds”, Leonardo Electronic Journal of Practices and Technologies, iss.11, pp.71-80, December.
[36] Birkhold, M., C. Friedrich, and A. Lechler, 2015, “Automation of the Casting Process using a model-based NC Architecture”, International Federation of Automatic Control-PapersOnLine, vol.48(17), pp.195–200, August. doi.org/10.1016/j.ifacol.2015.10.102
[37] CMU Panoptic Dataset. Retrieve from http://domedb.perception.cs.cmu.edu/
[38] Rother, C., V. Kolmogorov, and A. Blake, 2004, “"GrabCut": interactive foreground extraction using iterated graph cuts”, Association for Computing Machinery Transactions on Graphics, vol.23(3), pp. 309–314, August. doi.org/10.1145/1015706.1015720
[39] Kalal, Z., K. Mikolajczyk and J. Matas, 2012, “Tracking-Learning-Detection”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 34(7), pp. 1409-1422, July. doi.org/10.1109/TPAMI.2011.239
[40] Kazemi, V. and J. Sullivan, 2014, “One millisecond face alignment with an ensemble of regression trees”, In 2014 IEEE Conference on Computer Vision and Pattern Recognition, pp. 1867-1874, Columbus, Ohio, USA, 23-28 June 2014. doi.org/10.1109/CVPR.2014.241
[41] Grabner, H., M. Grabner and H. Bischof, “Real-Time Tracking via On-line Boosting”,2006, In Proceedings of the British Machine Vision Conference, pp.6.1-6.10, Edinburgh, 4-7 September 2006. doi.org/10.5244/C.20.6
[42] Babenko, B., M. Yang and S. Belongie, 2009, “Visual tracking with online Multiple Instance Learning”, In 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 983-990, Miami Florida, 20-25 June 2009, doi.org/10.1109/CVPR.2009.5206737
[43] Held, D., S. Thrun, and S. Savarese, 2016, “Learning to Track at 100 FPS with Deep Regression Networks”, In Computer Vision – ECCV 2016, pp.749-765, Amsterdam, The Netherlands, 11-14 October 2016. doi.org/10.1007/978-3-319-46448-0_45
[44] Henriques, J. F., R. Caseiro, P. Martins and J. Batista, 2015, “High-Speed Tracking with Kernelized Correlation Filters”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 37(3), pp. 583-596, March. doi.org/10.1109/TPAMI.2014.2345390
[45] Kalal, Z., K. Mikolajczyk, and J. Matas, 2010, “Forward-Backward Error: Automatic Detection of Tracking Failures”, In The 2010 20th International Conference on Pattern Recognition, pp. 2756-2759, N.W. Washington, D.C U.S., 23-26 August 2010. doi.org/10.1109/ICPR.2010.675
[46] Dalal, N., and B. Triggs, 2005, “Histograms of oriented gradients for human detection”, In 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), vol. 1, pp. 886-893, San Diego, CA, USA, 20-25 July 2005. doi.org/10.1109/CVPR.2005.177
[47] Ogino, Y., M. Yamagata, and N. Ichihashi, 2020, “Verification of Reliability and Validity of Motion Analysis Systems During Bilateral Squat Using Human Pose Tracking Algorithm.” Gait Posture. vol.80, pp.62-67, July. doi.org/10.1016/j.gaitpost.2020.05.027
[48] Nakano, N., T. Sakura, K. Ueda, L. Omura, A. Kimura, Y. Iino, S. Fukashiro, and S. Yoshioka, 2020, “Evaluation of 3D Markerless Motion Capture Accuracy Using OpenPose With Multiple Video Cameras”, Frontiers in Sports and Active Living, 2:50, May. doi.org/10.3389/fspor.2020.00050
[49] Takeda, I., A. Yamada, and H. Onodera, 2020, “Artificial Intelligence-Assisted Motion Capture for Medical Applications: A Comparative Study between Markerless and Passive Marker Motion Capture”, Computer Methods in Biomechanics and Biomedical Engineering, vol.24(8), pp.864-873, August. doi.org/10.1080/10255842.2020.1856372
[50] Zago, M., M. Luzzago, T. Marangoni, C. M. De, M. Tarabini, and M. Galli, 2020, “3D Tracking of Human Motion Using Visual Skeletonization and Stereoscopic Vision”, Frontiers in Bioengineering and Biotechnology. 8:181, March. doi.org/10.3389/fbioe.2020.00181