Semi-Supervised 3D Human Pose Estimation by Jointly Considering Temporal and Multiview Information

Chu, Wei-Ta; Pan, Zong-Wei

doi:10.1109/access.2020.3045794

Cited by 6 publications

(3 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This category of 3D-HPE works tries to learn more robust 3D pose estimators by combining annotated 3D pose training data with much more unlabeled video data. Existing works [ 40 , 41 , 42 ] have employed a dual-branch training pipeline with a fully supervised branch and a self-supervised branch that learns from the 2D pose inputs without 3D pose annotations. Rhodin et al [ 43 ] proposed addressing the problem of insufficiently large training samples by learning a latent representation of 3D geometry from multi-view 2D images.…”

Section: Related Workmentioning

confidence: 99%

PosturePose: Optimized Posture Analysis for Semi-Supervised Monocular 3D Human Pose Estimation

Amadi,

Agam

2023

Sensors

View full text Add to dashboard Cite

One motivation for studying semi-supervised techniques for human pose estimation is to compensate for the lack of variety in curated 3D human pose datasets by combining labeled 3D pose data with readily available unlabeled video data—effectively, leveraging the annotations of the former and the rich variety of the latter to train more robust pose estimators. In this paper, we propose a novel, fully differentiable posture consistency loss that is unaffected by camera orientation and improves monocular human pose estimators trained with limited labeled 3D pose data. Our semi-supervised monocular 3D pose framework combines biomechanical pose regularization with a multi-view posture (and pose) consistency objective function. We show that posture optimization was effective at decreasing pose estimation errors when applied to a 2D–3D lifting network (VPose3D) and two well-studied datasets (H36M and 3DHP). Specifically, the proposed semi-supervised framework with multi-view posture and pose loss lowered the mean per-joint position error (MPJPE) of leading semi-supervised methods by up to 15% (−7.6 mm) when camera parameters of unlabeled poses were provided. Without camera parameters, our semi-supervised framework with posture loss improved semi-supervised state-of-the-art methods by 17% (−15.6 mm decrease in MPJPE). Overall, our pose models compete favorably with other high-performing pose models trained under similar conditions with limited labeled data.

show abstract

Section: Related Workmentioning

confidence: 99%

PosturePose: Optimized Posture Analysis for Semi-Supervised Monocular 3D Human Pose Estimation

Amadi,

Agam

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…The prospect of learning more robust and accurate 3D pose estimators by combining vastly available unlabelled video data with limited annotated 3D pose datasets motivates the study of weakly-supervised HPE. Existing works [3,1,12,13] employ a dual-branch architecture with a fullysupervised branch and a semi-supervised branch that learns from 2D data without 3D pose annotations in a limited data scenario. Rhodin et al [14] propose to address the problem of insufficiently large training samples by learning a latent representation of 3D geometry from multi-view 2D images.…”

Section: Related Workmentioning

confidence: 99%

Boosting the Performance of Weakly-Supervised 3d Human Pose Estimators With Pose Prior Regularizers

Amadi

Agam

2022

2022 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

This work aims to boost the performance of 3D human pose estimators trained in a weakly-supervised setting where there are much fewer annotated 3D poses than unlabeled video data. We formulate two self-supervised pose prior regularizers (PPR) -bone proportion and joint mobility constraintsthat are pose translation, scale, and rotation invariant. These regularizers, combined with bone symmetry loss, reduce overfitting to the 2D reprojection loss commonly used in weakly-supervised settings by optimizing the bone lengths and joint rotations of estimated 3D poses. Consequently, improving the accuracy of 3D pose estimators. The regularizers are network independent and can be applied to any network architecture without modifications. We apply our proposed PPR to VideoPose3D network [1] and show that it decreases the MPJPE by 24% when using ≤5% of annotated H36M [2] 3D data, improving state-of-the-art accuracy by 7.9 mm.

show abstract

“…To date, with the advanced pose estimation algorithms, 2D or even 3D skeleton can be obtained from 2D images [5][6][7]. Although skeleton position can be obtained, feature extraction is required for further analysis.…”

Section: Introductionmentioning

confidence: 99%

Center Of Mass Trajectory: An Image Descriptor For Baseball Swing Analysis Based On Single Low-Cost Camera

Fang

Hsu

Chiu

et al. 2021

2021 IEEE International Conference on Multimedia &Amp; Expo Workshops (ICMEW)

View full text Add to dashboard Cite

This paper examined the plausibility of using 2D images to estimate the center of mass of the human body (CoMb) and the performance using the trajectory of CoMb as an event detector for baseball swing movement. Movement data from three subjects are collected using both a motion capture system and a commercial camera, and the resultant 2D skeleton and CoMb position from the 2D images are obtained using two pose estimation algorithms and an anthropometry model. The errors of the 2D results are 42.67 mm for the skeleton and 50.69 mm for CoMb estimation. 88 baseball swing movement videos are used to illustrate the performance of CoMb trajectory as an event detector with a baseball expert. The agreement was 90.91% between the CoMb trajectory and the baseball expert.

show abstract

Semi-Supervised 3D Human Pose Estimation by Jointly Considering Temporal and Multiview Information

Cited by 6 publications

References 27 publications

PosturePose: Optimized Posture Analysis for Semi-Supervised Monocular 3D Human Pose Estimation

PosturePose: Optimized Posture Analysis for Semi-Supervised Monocular 3D Human Pose Estimation

Boosting the Performance of Weakly-Supervised 3d Human Pose Estimators With Pose Prior Regularizers

Center Of Mass Trajectory: An Image Descriptor For Baseball Swing Analysis Based On Single Low-Cost Camera

Contact Info

Product

Resources

About