Filters








49,583 Hits in 6.5 sec

Whole-Body Human Pose Estimation in the Wild [article]

Sheng Jin, Lumin Xu, Jin Xu, Can Wang, Wentao Liu, Chen Qian, Wanli Ouyang, Ping Luo
2020 arXiv   pre-print
This paper investigates the task of 2D human whole-body pose estimation, which aims to localize dense landmarks on the entire human body including face, hands, body, and feet.  ...  To fill in this blank, we introduce COCO-WholeBody which extends COCO dataset with whole-body annotations.  ...  ZoomNet: Whole-Body Pose Estimation In this section, we will introduce our whole-body pose estimation pipeline.  ... 
arXiv:2007.11858v1 fatcat:rwjd2cwbovhgvd5ifz3cqw2ihy

DOPE: Distillation Of Part Experts for whole-body 3D pose estimation in the wild [article]

Philippe Weinzaepfel, Romain Brégier, Hadrien Combaluzier, Vincent Leroy, Grégory Rogez
2020 arXiv   pre-print
We introduce DOPE, the first method to detect and estimate whole-body 3D human poses, including bodies, hands and faces, in the wild.  ...  The main challenge is the lack of in-the-wild data with labeled whole-body 3D poses.  ...  Conclusion We have proposed DOPE, the first learning-based method to detect and estimate whole-body 3D human poses in the wild, including body, hand and face 2D-3D keypoints.  ... 
arXiv:2008.09457v1 fatcat:t7w5vcg2bzg5vc2fkgysyyst6y

Synthetic Training for Monocular Human Mesh Recovery [article]

Yu Sun and Qian Bao and Wu Liu and Wenpeng Gao and Yili Fu and Chuang Gan and Tao Mei
2020 arXiv   pre-print
Besides, to strengthen the generalization ability, most existing methods have used in-the-wild 2D pose datasets to supervise the estimated 3D pose via 3D-to-2D projection.  ...  This paper aims to estimate 3D mesh of multiple body parts (e.g., body, hands) with large-scale differences from a single RGB image.  ...  Specifically, for the pose-related estimation in both branches, we employ the whole-body 2D pose predicted from the 2D image as an intermediate representation.  ... 
arXiv:2010.14036v1 fatcat:7hqbnwgnufeyhlljuzhq7y3grm

A Review on Human Pose Estimation [article]

Rohit Josyula, Sarah Ostadabbas
2021 arXiv   pre-print
Therefore, we can define the problem of Human Pose Estimation as the localization of human joints or predefined landmarks in images and videos.  ...  The phenomenon of Human Pose Estimation (HPE) is a problem that has been explored over the years, particularly in computer vision. But what exactly is it?  ...  HPE in the Wild [16] Body 23 BlazePose [17] Body 33 Whole-Body HPE in the Wild [16] Face 68 CNN Facial Landmark Github [18] Face 68 Facial Keypoint Detection Github [19] Face 68 MediaPipe  ... 
arXiv:2110.06877v1 fatcat:b3n3uqyuavc4zcf54a4rydlf7m

FrankMocap: Fast Monocular 3D Hand and Body Motion Capture by Regression and Integration [article]

Yu Rong, Takaaki Shiratori, Hanbyul Joo
2020 arXiv   pre-print
In this paper, we present FrankMocap, a motion capture system that can estimate both 3D hand and body motion from in-the-wild monocular inputs with faster speed (9.5 fps) and better accuracy than previous  ...  Our method aims to capture 3D body and hand motion simultaneously from challenging in-the-wild monocular videos.  ...  We also want to thank Xintao Wang for helping us in implementing motion blur augmentation.  ... 
arXiv:2008.08324v1 fatcat:e7uhjzwx6neqvka6matnhzp5u4

FrankMocap: A Monocular 3D Whole-Body Pose Estimation System via Regression and Integration [article]

Yu Rong, Takaaki Shiratori, Hanbyul Joo
2021 arXiv   pre-print
In this paper, we present FrankMocap, a fast and accurate whole-body 3D pose estimation system that can produce 3D face, hands, and body simultaneously from in-the-wild monocular images.  ...  Most existing monocular 3D pose estimation approaches only focus on a single body part, neglecting the fact that the essential nuance of human motion is conveyed through a concert of subtle movements of  ...  Input Hand-Only Whole Body Side View Figure 16 : Qualitative results for FrankMocap with wrist integration network adopted.  ... 
arXiv:2108.06428v1 fatcat:e7qbykmj5ndnxctskx42yafd3i

NeuralAnnot: Neural Annotator for 3D Human Mesh Training Sets [article]

Gyeongsik Moon and Hongsuk Choi and Kyoung Mu Lee
2022 arXiv   pre-print
The 3D pseudo-GTs are obtained by annotators, systems that iteratively fit 3D human model parameters to GT 2D/3D joint coordinates of training sets in the pre-processing stage of the regressors.  ...  Most 3D human mesh regressors are fully supervised with 3D pseudo-GT human model parameters and weakly supervised with GT 2D/3D joint coordinates as the 3D pseudo-GTs bring great performance gain.  ...  Recently, EFT [12] is introduced to obtain pseudo-GT 3D human body pose and mesh from in-the-wild images.  ... 
arXiv:2011.11232v5 fatcat:nwt2mvt7szbspmfcszxjjgsfxa

VIBE: Video Inference for Human Body Pose and Shape Estimation [article]

Muhammed Kocabas, Nikos Athanasiou, Michael J. Black
2020 arXiv   pre-print
To address this problem, we propose Video Inference for Body Pose and Shape Estimation (VIBE), which makes use of an existing large-scale motion capture dataset (AMASS) together with unpaired, in-the-wild  ...  We define a temporal network architecture and show that adversarial training, at the sequence level, produces kinematically plausible motion sequences without in-the-wild ground-truth 3D labels.  ...  This research was partially supported by the Max Planck ETH Center for Learning Systems and the Max Planck Graduate Center for Computer and Information Science.  ... 
arXiv:1912.05656v3 fatcat:vrhvro6q65ftzpuzehb4ab24bu

[Invited Paper] Fast and Accurate Whole-Body Pose Estimation in the Wild and Its Applications

Jianfeng XU, Kazuyuki TASAKA, Masashi YAMAGUCHI
2021 ITE Transactions on Media Technology and Applications  
Here, we present a fast and accurate in-the-wild whole-body pose estimation system. Our system detects not only body keypoints, but also foot and hand keypoints, with high accuracy in real time.  ...  Recently, multi-person pose estimation techniques have drawn significant attention in both academia and industry due to their great potential utility.  ...  AP@0.75 1.0 69.7% 88.3% 77.0% 1.3 69.6% 88.3% 76.8% 1.5 68.2% 87.9% 76.2% Paper » Fast and Accurate Whole-Body Pose Estimation in the Wild and Its Applications  ... 
doi:10.3169/mta.9.63 fatcat:aals3oyr6rdwbjcypddyrdiebq

Single-Network Whole-Body Pose Estimation [article]

Gines Hidalgo and Yaadhav Raaj and Haroon Idrees and Donglai Xiang and Hanbyul Joo and Tomas Simon and Yaser Sheikh
2019 arXiv   pre-print
Our approach considerably improves upon OpenPose , the only work so far capable of whole-body pose estimation, both in terms of speed and global accuracy.  ...  We present the first single-network approach for 2D whole-body pose estimation, which entails simultaneous localization of body, face, hands, and feet keypoints.  ...  Whole-Body Pose Estimation We want whole-body pose estimation to be accurate but also fast.  ... 
arXiv:1909.13423v1 fatcat:gswjhtdvkzf6ziorry6owakmfa

Post-Data Augmentation to Improve Deep Pose Estimation of Extreme and Wild Motions [article]

Kohei Toyoda, Michinari Kono, Jun Rekimoto
2019 arXiv   pre-print
One of the commonly used DNNs is human pose estimation. This kind of technique is widely used for motion capturing of humans, and to generate or modify virtual avatars.  ...  To address these issues, we propose a method to improve the pose estimation accuracy for extreme/wild motions by using pre-trained models, i.e., without performing the training procedure by yourselves.  ...  ACKNOWLEDGMENTS The work is supported by ISID Technosolutions, Ltd.  ... 
arXiv:1902.04250v1 fatcat:njbsd6p6xzeevmn3i7vah25nqa

DensePose: Dense Human Pose Estimation In The Wild [article]

Rıza Alp Güler, Natalia Neverova, Iasonas Kokkinos
2018 arXiv   pre-print
In this work, we establish dense correspondences between RGB image and a surface-based representation of the human body, a task we refer to as dense human pose estimation.  ...  We then use our dataset to train CNN-based systems that deliver dense correspondence 'in the wild', namely in the presence of background, occlusions and scale variations.  ...  Acknowledgements We thank the authors of [15] for sharing their code, Piotr Dollar for guidance and proposals related to our dataset's quality, Tsung-Yi Lin for his help with COCO-related issues and  ... 
arXiv:1802.00434v1 fatcat:34st44efsnbnbn5x4wxlxl3cv4

DensePose: Dense Human Pose Estimation in the Wild

Riza Alp Guler, Natalia Neverova, Iasonas Kokkinos
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
Figure 1 : Dense pose estimation aims at mapping all human pixels of an RGB image to the 3D surface of the human body.  ...  Abstract In this work we establish dense correspondences between an RGB image and a surface-based representation of the human body, a task we refer to as dense human pose estimation.  ...  Acknowledgements We thank the authors of [16] for their code, P. Dollar and T.-Y. Lin for help with COCO, the authors of [28] for making the SMPL model open for research and H. Y.  ... 
doi:10.1109/cvpr.2018.00762 dblp:conf/cvpr/GulerNK18 fatcat:tcemagmzprctbdjdk3325kzeta

Pose2Mesh: Graph Convolutional Network for 3D Human Pose and Mesh Recovery from a 2D Human Pose [article]

Hongsuk Choi, Gyeongsik Moon, Kyoung Mu Lee
2021 arXiv   pre-print
The 2D human pose as input provides essential human body articulation information, while having a relatively homogeneous geometric property between the two domains.  ...  Most of the recent deep learning-based 3D human pose and mesh estimation methods regress the pose and shape parameters of human mesh models, such as SMPL and MANO, from an input image.  ...  Related works 3D human body pose estimation.  ... 
arXiv:2008.09047v3 fatcat:5z5d7ok67vgnpipmuqqqqg2p3i

Delving Deep Into Hybrid Annotations for 3D Human Recovery in the Wild [article]

Yu Rong, Ziwei Liu, Cheng Li, Kaidi Cao, Chen Change Loy
2019 arXiv   pre-print
Though much progress has been achieved in single-image 3D human recovery, estimating 3D model for in-the-wild images remains a formidable challenge.  ...  Specifically, we focus on the challenging task of in-the-wild 3D human recovery from single images when paired 3D annotations are not fully available.  ...  Dense correspondence, namely, IUV maps generated by Dense-Pose [1, 19] , is an effective annotations for in-the-wild 3D human recovery.  ... 
arXiv:1908.06442v2 fatcat:6imlyumwyfcdzjloi5cej4tloy
« Previous Showing results 1 — 15 out of 49,583 results