The feasibility and validity of a single camera deep learning-based method for 3D biomechanical analysis in strength training
proof-of-concept
DOI:
https://doi.org/10.51224/SRXIV.206Keywords:
Strength training, Video-based motion capture, Marker-less motion capture, Biomechanics, Deep-learningAbstract
Biomechanical analysis is valuable for injury risk and performance assessment in sports, but the application is limited due to restrictions in costs, set-up time and accuracy of available motion capture methods. Therefore, the present proof-of-concept study evaluated the feasibility and validity of a more suitable method, based on a single camera combined with a deep-learning algorithm, by comparing obtained biomechanical parameters with those obtained by the state-of-the-art optoelectronic measurement system (OMS) and the marker-less Kinect during upper-body strength exercises. Results from five athletes revealed strong to excellent correlations for most parameters and root-mean-square deviations of 4-8 degrees for angles and 0.9-1.4Nm for moments, but insufficient ICCs compared to the OMS, and partly better performance than the Kinect. In conclusion, the present study showed that the single camera deep learning-based method is feasible for biomechanical analysis of strength exercises and provides limited evidence that some parameters can be estimated with reasonable accuracy. However, the accuracy of peak angle and moment estimations should be improved before this method can be applied for injury prevention, i.e. by training the deep-learning model on a larger variety of subject anthropometries. Furthermore, future research should investigate the validity for larger sample sizes and multiple exercises.
Metrics
References
Bini, R.R. and F.P. Carpes, Introduction to biomechanical analysis for performance enhancement and injury prevention, in Biomechanics of Cycling. 2014, Springer. p. 1-11. doi: http://dx.doi.org/10.1007/978-3-319-05539-8_1
van der Kruk, E. and M.M. Reijne, Accuracy of human motion capture systems for sport applications; state-of-the-art review. European journal of sport science, 2018. 18(6): p. 806-819. doi: http://dx.doi.org/10.1080/17461391.2018.1463397
Stagni, R., et al., Quantification of soft tissue artefact in motion analysis by combining 3D fluoroscopy and stereophotogrammetry: a study on two subjects. Clinical Biomechanics, 2005. 20(3): p. 320-329. doi: http://dx.doi.org/10.1016/j.clinbiomech.2004.11.012
Bonnechere, B., et al., Validity and reliability of the Kinect within functional assessment activities: comparison with standard stereophotogrammetry. Gait & posture, 2014. 39(1): p. 593-598. doi: http://dx.doi.org/10.1016/j.gaitpost.2013.09.018
Begon, M., et al., Computation of the 3D kinematics in a global frame over a 40 m-long pathway using a rolling motion analysis system. Journal of biomechanics, 2009. 42(16): p. 2649-2653.
Skals, S., et al., A musculoskeletal model driven by dual Microsoft Kinect Sensor data. Multibody System Dynamics, 2017. 41(4): p. 297-316. doi: http://dx.doi.org/10.1007/s11044-017-9573-8
Plantard, P., et al., Pose estimation with a kinect for ergonomic studies: Evaluation of the accuracy using a virtual mannequin. Sensors, 2015. 15(1): p. 1785-1803. doi: http://dx.doi.org/10.3390/s150101785
Dutta, T., Evaluation of the Kinect™ sensor for 3-D kinematic measurement in the workplace. Applied ergonomics, 2012. 43(4): p. 645-649. doi: http://dx.doi.org/10.1016/j.apergo.2011.09.011
Pavllo, D., et al. 3d human pose estimation in video with temporal convolutions and semi-supervised training. in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2019. doi: http://dx.doi.org/10.1109/CVPR.2019.00794
Luvizon, D.C., D. Picard, and H. Tabia. 2d/3d pose estimation and action recognition using multitask deep learning. in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2018. doi: http://dx.doi.org/10.1109/CVPR.2018.00539
Wade, L., et al., Applications and limitations of current markerless motion capture methods for clinical gait biomechanics. PeerJ, 2022. 10: p. E12995. doi: http://dx.doi.org/10.7717/peerj.12995
Spiers, A., S. Osborne, and M. Li, Engagement in Fitness Activities in England, in The Rise and Size of the Fitness Industry in Europe. 2020, Springer. p. 157-175. doi:
Kemler, E., H. Valkenberg, and E. Verhagen, More people more active, but there is a counter site. Novice athletes are at highest risk of injury in a large population-based retrospective cross-sectional study. BMJ Open Sport & Exercise Medicine, 2022. 8(1): p. E001255. doi: http://dx.doi.org/10.1136/bmjsem-2021-001255
Shotton, J., et al. Real-time human pose recognition in parts from single depth images. in CVPR 2011. 2011. Ieee. doi: http://dx.doi.org/10.1109/CVPR.2011.5995316
Pavllo, D., et al. 2019; https://github.com/facebookresearch/VideoPose3D.
Pavllo, D., et al. 2019; https://github.com/facebookresearch/VideoPose3D/blob/main/INFERENCE.md
Wu, G., et al., ISB recommendation on definitions of joint coordinate systems of various joints for the reporting of human joint motion—Part II: shoulder, elbow, wrist and hand. Journal of biomechanics, 2005. 38(5): p. 981-992. doi: http://dx.doi.org/10.1016/j.jbiomech.2004.05.042
Plantard, P., et al., Inverse dynamics based on occlusion-resistant Kinect data: Is it usable for ergonomics? International Journal of Industrial Ergonomics, 2017. 61: p. 71-80. doi: http://dx.doi.org/10.1016/j.ergon.2017.05.010
Hof, A.L., An explicit expression for the moment in multibody systems. Journal of biomechanics, 1992. 25(10): p. 1209-1211. doi: http://dx.doi.org/10.1016/0021-9290(92)90076-D
Molenbroek, J., DINED, Anthropometric database. Delft University of Technology, 2004. doi:
De Leva, P., Adjustments to Zatsiorsky-Seluyanov's segment inertia parameters. Journal of biomechanics, 1996. 29(9): p. 1223-1230. doi: http://dx.doi.org/10.1016/0021-9290(95)00178-6
Taylor, R., Interpretation of the correlation coefficient: a basic review. Journal of diagnostic medical sonography, 1990. 6(1): p. 35-39. doi: http://dx.doi.org/10.1177/875647939000600106
Nunnally, J.C. and I.H. Bernstein, Psychometric theory. 1994. doi:
Zaki, R., et al., Statistical methods used to test for agreement of medical instruments measuring continuous variables in method comparison studies: a systematic review. PloS one, 2012. 7(5): p. E37908. doi: http://dx.doi.org/10.1371/journal.pone.0037908
Chen, X., et al. Hand pose estimation in depth image using CNN and random forest. in MIPPR 2017: Pattern Recognition and Computer Vision. 2018. International Society for Optics and Photonics. doi: http://dx.doi.org/10.1117/12.2288114
Norman, R., et al., A comparison of peak vs cumulative physical work exposure risk factors for the reporting of low back pain in the automotive industry. Clinical biomechanics, 1998. 13(8): p. 561-573. doi: http://dx.doi.org/10.1016/S0268-0033(98)00020-5
Hansson, G.-Å., et al., Physical workload in various types of work: Part II. Neck, shoulder and upper arm. International Journal of Industrial Ergonomics, 2010. 40(3): p. 267-281. doi: http://dx.doi.org/10.1016/j.ergon.2009.11.002
Downloads
Posted
License
Copyright (c) 2022 Lisa Noteboom, Marco Hoozemans, DirkJan Veeger, Frans van der Helm
This work is licensed under a Creative Commons Attribution 4.0 International License.
