1Paulo V. Opiña Jr., 2Arnel C. Fajardo
1,2Isabela State University
DOI : https://doi.org/10.47191/ijmra/v7-i04-44Google Scholar Download Pdf
ABSTRACT:
Human pose estimation (HPE) is a critical computer vision task that underpins various applications such as motion analysis and human-computer interaction. Deep learning has significantly improved HPE accuracy, but challenges, specifically in limited training data and occlusions remain. This work focuses on real-time HPE on mobile devices. We compare three prominent frameworks (BlazePose, OpenPose, AlphaPose) by analyzing their input data requirements, underlying inference procedures, and performance on popular datasets. We also explore commonly used HPE benchmark datasets and evaluation metrics. This comparative analysis provides a clear understanding of the current state-of-the-art for real-time mobile HPE, benefiting researchers and developers working on mobile applications that leverage HPE functionalities.
KEYWORDS:Computer Vision, Human Pose Estimation, OpenPose, AlphaPose, BlazePose
REFERENCES1)C. Zheng et al., “Deep Learning-Based Human Pose Estimation: A Survey,” IEEE Comput. Vis. Pattern Recognit., Dec. 2020, [Online]. Available: http://arxiv.org/abs/2012.13392
2) V. Bazarevsky, I. Grishchenko, K. Raveendran, T. Zhu, F. Zhang, and M. Grundmann, “BlazePose: On-device Real-time Body Pose tracking,” IEEE Comput. Vis. Pattern Recognit., Jun. 2020, [Online]. Available: http://arxiv.org/abs/2006.10204
3) Z. Cao, G. Hidalgo, T. Simon, S.-E. Wei, and Y. Sheikh, “OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields,” IEEE Comput. Vis. Pattern Recognit., Dec. 2018, [Online]. Available: http://arxiv.org/abs/1812.08008
4) H.-S. Fang et al., “AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Time,” IEEE Comput. Vis. Pattern Recognit., Nov. 2022, [Online]. Available: http://arxiv.org/abs/2211.03375
5) T. L. Munea, Y. Z. Jembre, H. T. Weldegebriel, L. Chen, C. Huang, and C. Yang, “The Progress of Human Pose Estimation: A Survey and Taxonomy of Models Applied in 2D Human Pose Estimation,” IEEE Access, vol. 8, pp. 133330–133348, 2020, doi: 10.1109/ACCESS.2020.3010248.
6) Z. D. Ce Zheng, Sijie Zhu, Matias Mendieta, Taojiannan Yang, Chen Chen, “3D Human Pose Estimation with Spatial and Temporal Transformers,” IEEE Comput. Vis. Pattern Recognit., 2021.
7) J. Wang et al., “Deep 3D human pose estimation: A review,” IEEE Comput. Vis. Pattern Recognit., vol. 210, p. 103225, Sep. 2021, doi: 10.1016/j.cviu.2021.103225.
8) Y. C. Cong Yu, Dongheng Zhang, Zhi Wu, Zhi Lu, Chunyang Xie, Yang Hu, “RFPose-OT: RF-Based 3D Human Pose Estimation via Optimal Transport Theory,” IEEE Access, 2022.
9) M. Tölgyessy, M. Dekan, Ľ. Chovanec, and P. Hubinský, “Evaluation of the Azure Kinect and Its Comparison to Kinect V1 and Kinect V2,” IEEE Comput. Vis. Pattern Recognit., vol. 21, no. 2, p. 413, Jan. 2021, doi: 10.3390/s21020413.
10) A. Toshev, “Deeppose: Human pose estimation via deep neural networks,” IEEE Comput. Vis. Pattern Recognit., 2014.
11) Z. Li, F. Liu, W. Yang, S. Peng, and J. Zhou, “A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects,” IEEE Trans. Neural Networks Learn. Syst., vol. 33, no. 12, pp. 6999–7019, Dec. 2022, doi: 10.1109/TNNLS.2021.3084827.
12) B. Koonce, “ResNet 50,” in IEEE on Computer Vision and Pattern Recognition, Berkeley, CA: Apress, 2021, pp. 63–72. doi: 10.1007/978-1-4842-6168-2_6.
13) L. Dinh, “Density estimation using Real NVP,” IEEE Comput. Vis. Pattern Recognit., 2017.
14) S.-E. Wei, “Convolutional Pose Machines,” IEEE Comput. Vis. Pattern Recognit., 2016.
15) A. Newell, K. Yang, and J. Deng, “Stacked Hourglass Networks for Human Pose Estimation,” in IEEE on Computer Vision and Pattern Recognition, 2016, pp. 483–499. doi: 10.1007/978-3-319-46484-8_29.
16) D. Han, “Deep Pyramidal Residual Networks,” IEEE Comput. Vis. Pattern Recognit., 2017.
17) J. Wang, “Deep High-Resolution Representation Learning for Visual Recognition,” IEEE Comput. Vis. Pattern Recognit., 2020.
18) A. Creswell, T. White, V. Dumoulin, K. Arulkumaran, B. Sengupta, and A. A. Bharath, “Generative Adversarial Networks: An Overview,” IEEE Signal Process. Mag., vol. 35, no. 1, pp. 53–65, Jan. 2018, doi: 10.1109/MSP.2017.2765202.
19) Z. Cao, “Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields,” IEEE Comput. Vis. Pattern Recognit., 2017.
20) J. Redmon, “YOLOv3: An Incremental Improvement,” IEEE Comput. Vis. Pattern Recognit., 2018, doi: https://doi.org/10.48550/arXiv.1804.02767.
Volume 07 Issue 04 April 2024
There is an Open Access article, distributed under the term of the Creative Commons Attribution – Non Commercial 4.0 International (CC BY-NC 4.0) (https://creativecommons.org/licenses/by-nc/4.0/), which permits remixing, adapting and building upon the work for non-commercial use, provided the original work is properly cited.
Our Services and Policies
Authors should prepare their manuscripts according to the instructions given in the authors' guidelines. Manuscripts which do not conform to the format and style of the Journal may be returned to the authors for revision or rejected.
The Journal reserves the right to make any further formal changes and language corrections necessary in a manuscript accepted for publication so that it conforms to the formatting requirements of the Journal.
International Journal of Multidisciplinary Research and Analysis will publish 12 monthly online issues per year,IJMRA publishes articles as soon as the final copy-edited version is approved. IJMRA publishes articles and review papers of all subjects area.
Open access is a mechanism by which research outputs are distributed online, Hybrid open access journals, contain a mixture of open access articles and closed access articles.
International Journal of Multidisciplinary Research and Analysis initiate a call for research paper for Volume 07 Issue 11 (November 2024).
PUBLICATION DATES:
1) Last Date of Submission : 26 November 2024 .
2) Article published within a week.
3) Submit Article : editor@ijmra.in or Online
Why with us
1 : IJMRA only accepts original and high quality research and technical papers.
2 : Paper will publish immediately in current issue after registration.
3 : Authors can download their full papers at any time with digital certificate.
The Editors reserve the right to reject papers without sending them out for review.
Authors should prepare their manuscripts according to the instructions given in the authors' guidelines. Manuscripts which do not conform to the format and style of the Journal may be returned to the authors for revision or rejected. The Journal reserves the right to make any further formal changes and language corrections necessary in a manuscript accepted for publication so that it conforms to the formatting requirements of the Journal.