2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2021
DOI: 10.1109/cvpr46437.2021.00708
|View full text |Cite
|
Sign up to set email alerts
|

SimPoE: Simulated Character Control for 3D Human Pose Estimation

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
63
0

Year Published

2021
2021
2022
2022

Publication Types

Select...
5
3

Relationship

1
7

Authors

Journals

citations
Cited by 110 publications
(64 citation statements)
references
References 43 publications
0
63
0
Order By: Relevance
“…For invisible poses, since there can be many plausible poses beside the GT, we follow prior work [3,102] to compute the best PA-MPJPE out of multiple samples for our probabilistic approach. (3) Accel, which computes the mean acceleration error of each joint and is commonly used to measure the jitter in estimated motions [45,104]. ( 4) FID, which is an extension of the original Frechet Inception Distance that calculates the distribution distance between estimated motions and the GT.…”
Section: Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…For invisible poses, since there can be many plausible poses beside the GT, we follow prior work [3,102] to compute the best PA-MPJPE out of multiple samples for our probabilistic approach. (3) Accel, which computes the mean acceleration error of each joint and is commonly used to measure the jitter in estimated motions [45,104]. ( 4) FID, which is an extension of the original Frechet Inception Distance that calculates the distribution distance between estimated motions and the GT.…”
Section: Methodsmentioning
confidence: 99%
“…A few methods exploit various scene constraints during the optimization process to improve depth prediction [95,106]. Alternatively, recent approaches use physics-based constraints to ensure the physical plausibility of the estimated poses [12,34,84,98,104]. Iqbal et al [32] exploit a limblength constraint to recover the absolute translation of the person using a 2.5D representation.…”
Section: Related Workmentioning
confidence: 99%
“…In this way, the two networks can leverage local and global complementary information from the other one to obtain better representations. The high-level overview of FFB6D framework is shown in Fig 57 . Some of the other popular frameworks for human pose estimation includes: regional multi-person pose estimation [86], simple baselines for human pose estimation and tracking [87], OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields [82], SimPoE: Simulated Character Control for 3D Human Pose Estimation [88].…”
Section: Tracking and Pose Estimation For Armentioning
confidence: 99%
“…Physics-aware Inference Several recent works have introduced physical awareness to improve purely data-driven approaches [11,27,29,33,36,37,44]. [29] use a physics simulation to validate the plausibility of a generative model for objects via a stability measure.…”
Section: Related Workmentioning
confidence: 99%
“…To improve the task of human-pose reconstruction from videos, different methods have added physics-based modules to correct the output of a human-pose estimation models. This is achieved either in a post-processing optimization framework [33,37], with an apprixmation of physics [36], or via a reinforcement learning policy that directly corrects the pose estimate [44]. [27] regulate a data-driven policy for egocentric pose estimation with a physics-based policy.…”
Section: Related Workmentioning
confidence: 99%