Analyzing embedded pose estimation solutions for human behaviour understanding

• José Gomes da Silva Neto UFPE
• João Marcelo Xavier Natário Teixeira UFPE
• Veronica Teichrieb UFPE

DOI: https://doi.org/10.5753/svr_estendido.2020.12951

Resumo

This work represents the first phase of a more complete work that has the goal of using RGB images as information to make analyses of human behavior. In this phase, we developed a prototype of hardware/software, capable of estimating human pose using only RGB information. The equipment chosen was the NVIDIA Jetson Nano, known for having have a better computational performance compared to Raspberry pi and Arduino microcontoller alternatives. In the search for important algorithms for pose estimation, applied to the limited platform as the Jetson Nano, we found important works such as HyperPose, TensorRT Pose Estimation, and the used on the project, tf-poseestimation. The results show a low FPS performance of the Jetson Nano, using the chosen algorithm, compared to related hardware, such as the NVIDIA Jetson TX2 and NVIDIA Jetson Xavier.

Palavras-chave: Jetson Nano, RGB, Human pose Estimation, NVIDIA

Referências

Mykhaylo Andriluka, Stefan Roth, and Bernt Schiele. Pictorial structures revisited: People detection and articulated pose estimation. In 2009 IEEE conference on computer vision and pattern recognition, pages 1014–1021. IEEE, 2009.

Mykhaylo Andriluka, Stefan Roth, and Bernt Schiele. Monocular 3d pose estimation and tracking by detection. In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pages 623–630. IEEE, 2010.

Vasileios Belagiannis and Andrew Zisserman. Recurrent human pose estimation. In 2017 12th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2017), pages 468–475. IEEE, 2017.

Zhe Cao, Gines Hidalgo, Tomas Simon, Shih-En Wei, and Yaser Sheikh. Openpose: realtime multi-person 2d pose estimation using part affinity fields. arXiv preprint arXiv:1812.08008, 2018.

Pedro F Felzenszwalb and Daniel P Huttenlocher. Pictorial structures for object recognition. International journal of computer vision, 61(1):55–79, 2005.

Georgia Gkioxari, Bharath Hariharan, Ross Girshick, and Jitendra Malik. Using k-poselets for detecting people and localizing their keypoints. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 3582–3589, 2014.

Leonid Pishchulin, Mykhaylo Andriluka, Peter Gehler, and Bernt Schiele. Poselet conditioned pictorial structures. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 588–595, 2013.

Leonid Pishchulin, Arjun Jain, Mykhaylo Andriluka, Thorsten Thormahlen, and Bernt Schiele. Articulated people detection and pose estimation: Reshaping the future. In 2012 IEEE Conference on Computer Vision and Pattern Recognition, pages 3178–3185. IEEE, 2012.

Luiz José Schirmer Silva, Djalma Lúcio Soares da Silva, Alberto Barbosa Raposo, Luiz Velho, and Hélio Côrtes Vieira Lopes. Tensorpose: Real-time pose estimation for interactive applications. Computers & Graphics, 85:1–14, 2019.

Alexander Toshev and Christian Szegedy. Deeppose: Human pose estimation via deep neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 1653–1660, 2014.