A Spatial-temporal 3D Human Pose Reconstruction Framework
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
Three-dimensional (3D) human pose reconstruction from single-view image is a difficult and challenging topic. Existing approaches mostly process frame-by-frame independently while inter-frames are highly correlated ina sequence. In contrast, we introduce a novel spatial-temporal 3D human pose reconstruction framework thatleverages both intra and inter-frame relationships in consecutive 2D pose sequences. Orthogonal matchingpursuit (OMP) algorithm, pre-trained pose-angle limits and temporal models have been implemented. Severalquantitative comparisons between our proposed framework and recent works have been studied on CMUmotion capture dataset and Vietnamese traditional dance sequences. Our framework outperforms others by10% lower of Euclidean reconstruction error and more robust against Gaussian noise. Additionally, it is alsoimportant to mention that our reconstructed 3D pose sequences are more natural and smoother than others.
Abstract
Three-dimensional (3D) human pose reconstruction from single-view image is a difficult and challenging topic. Existing approaches mostly process frame-by-frame independently while inter-frames are highly correlated ina sequence. In contrast, we introduce a novel spatial-temporal 3D human pose reconstruction framework thatleverages both intra and inter-frame relationships in consecutive 2D pose sequences. Orthogonal matchingpursuit (OMP) algorithm, pre-trained pose-angle limits and temporal models have been implemented. Severalquantitative comparisons between our proposed framework and recent works have been studied on CMUmotion capture dataset and Vietnamese traditional dance sequences. Our framework outperforms others by10% lower of Euclidean reconstruction error and more robust against Gaussian noise. Additionally, it is alsoimportant to mention that our reconstructed 3D pose sequences are more natural and smoother than others.
참고문헌 (14)
- "Motion style extraction based on sparse coding decomposition" (2018)[URL]
- C. Sminchisescu, "Building roadmaps of local minima of visual models", Computer Vision–ECCV 2002, pp.566~582 (2002), Springer
- V. Ramakrishna, "Reconstructing 3D human pose from 2D image landmarks", Computer Vision–ECCV 2002, pp.573~586 (2012), Springer
- I. Akhter, "Pose-conditioned joint angle limits for 3D human pose reconstruction", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.1446~1455 (2015)
- C. Sminchisescu, "Estimating articulated human motion with covariance scaled sampling", The International Journal of Robotics Research, 22(6), pp.371~391 (2003)
- H. K. Kien, "Single view image based—3D human pose reconstruction", Proceedings of 2017 9th International Conference on Knowledge and Systems Engineering (KSE), pp.118~123 (2017)
- J. Chen, "Data-free prior model for upper body pose estimation and tracking", IEEE Transactions on Image Processing, 22(12), pp.4627~4639 (2013)
- J. M. Rehg, "Ambiguities in visual tracking of articulated objects using twoand three-dimensional models", The International Journal of Robotics Research, 22(6), pp.393~418 (2003)
- L. Pishchulin, "Deepcut:joint subset partition and labeling for multi person pose estimation", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.4929~4937 (2016)
- F. Bogo, "Keep it SMPL: automatic estimation of 3D human pose and shape from a single image", Computer Vision–ECCV 2016, pp.561~578 (2016)
- "Structured prediction of 3D human pose with deep neural networks" (2016)[URL]
- "CMU Graphics Lab Motion Capture Database"[URL]
- F. L. Bookstein, "Morphometric Tools for Landmark Data: Geometry and Biology" (1991), Cambridge University Press
- "YOLO9000: better, faster, stronger" (2016)[URL]