A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Filters
GraFormer: Graph Convolution Transformer for 3D Pose Estimation
[article]
2021
arXiv
pre-print
Preserved Fulltext
To alleviate this issue, we propose GraFormer, a novel transformer architecture combined with graph convolution for 3D pose estimation. ...
Exploiting relations among 2D joints plays a crucial role yet remains semi-developed in 2D-to-3D pose estimation. ...
We propose a novel model combined graph convolution and transformer, called GraFormer, for 3D pose estimation, which aims at better exploiting relations among graph-structured 2D joints. ...
arXiv:2109.08364v1
fatcat:2b2u6nkgtrgfhpft25p42qt44a
THOR-Net: End-to-end Graformer-based Realistic Two Hands and Object Reconstruction with Self-supervision
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Graph Convolutional networks (GCNs) allow for the preservation of the topologies of hands poses and shapes by modeling them as a graph. ...
The 3D poses of the hands and objects are reconstructed by the other branch using a GraFormer network. ...
GCNs for Pose Estimation Recently, 3D pose estimation from 2D pose using Graph Convolutional Networks (GCNs) showed very promising results [9, 43] . ...
arXiv:2210.13853v1
fatcat:f4vjngbnvfgydgax3pjvmdfe5q
Pose-Oriented Transformer with Uncertainty-Guided Refinement for 2D-to-3D Human Pose Estimation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
To tackle this issue, in this paper, we propose a Pose-Oriented Transformer (POT) with uncertainty guided refinement for 3D HPE. ...
There has been a recent surge of interest in introducing transformers to 3D human pose estimation (HPE) due to their powerful capabilities in modeling long-term dependencies. ...
Taking the feature extracted by CNNs as input, (Lin, Wang, and Liu 2021 ) further proposed a graph-convolution-reinforced transformer to predict 3D pose. ...
arXiv:2302.07408v1
fatcat:y63mddwzdfdp5lyhxfnos7bpte
K-Order Graph-oriented Transformer with GraAttention for 3D Pose and Shape Estimation
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We propose a novel attention-based 2D-to-3D pose estimation network for graph-structured data, named KOG-Transformer, and a 3D pose-to-shape estimation network for hand data, named GASE-Net. ...
By stacking GR-MSA and KOG-MSA, we propose a novel network KOG-Transformer for 2D-to-3D pose estimation. ...
First, we replace the KOG-MSA layers in KOG-Transformer with different graph convolutional networks used in the previous 2D-to-3D pose estimation methods for comparison. ...
arXiv:2208.11328v2
fatcat:6oretsp4hfaw5kjri2xalxqgfi
HopFIR: Hop-wise GraphFormer with Intragroup Joint Refinement for 3D Human Pose Estimation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2023 pre-print arXiv:2302.14581v1 fatcat:zfvaspu76fhurhoofjphxcfwqaOther Versions
2D-to-3D human pose lifting is fundamental for 3D human pose estimation (HPE), for which graph convolutional networks (GCNs) have proven inherently suitable for modeling the human skeletal topology. ...
The HGF module groups the joints by k-hop neighbors and applies a hopwise transformer-like attention mechanism to these groups to discover latent joint synergies. ...
Approaches in the second category decouple the 3D HPE task into 2D pose estimation from an image and 3D pose estimation from the detected 2D joints (2D-to-3D). For example, Martinez et al. ...
arXiv:2302.14581v3
fatcat:bvjgy6436jcqpdgwi4wh52f3ry
KTPFormer: Kinematics and Trajectory Prior Knowledge-Enhanced Transformer for 3D Human Pose Estimation
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Other Versions
This paper presents a novel Kinematics and Trajectory Prior Knowledge-Enhanced Transformer (KTPFormer), which overcomes the weakness in existing transformer-based methods for 3D human pose estimation that ...
More importantly, our KPA and TPA modules have lightweight plug-and-play designs and can be integrated into various transformer-based networks (i.e., diffusion-based) to improve the performance with only ...
Acknowledgments The work described in this paper is supported, in part, by the Innovation and Technology Commission of Hong Kong under grant ITP/028/21TP and by the Laboratory for Artificial Intelligence ...
arXiv:2404.00658v2
fatcat:lgvv2aidobbijf7ywhwdgdwtsm
Uncertainty-Aware Testing-Time Optimization for 3D Human Pose Estimation
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Specifically, during the training phase, we design an effective 2D-to-3D network for estimating the corresponding 3D pose while quantifying the uncertainty of each 3D joint. ...
Although data-driven methods have achieved success in 3D human pose estimation, they often suffer from domain gaps and exhibit limited generalization. ...
-3D lifting network for 3D human pose estimation. ...
arXiv:2402.02339v1
fatcat:ocb334d2qfhfbfepswhltc7d3u
HSTFormer: Hierarchical Spatial-Temporal Transformers for 3D Human Pose Estimation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Transformer-based approaches have been successfully proposed for 3D human pose estimation (HPE) from 2D pose sequence and achieved state-of-the-art (SOTA) performance. ...
To mitigate this issue, we propose Hierarchical Spatial-Temporal transFormers (HSTFormer) to capture multi-level joints' spatial-temporal correlations from local to global gradually for accurate 3D HPE ...
[1] utilize graph convolution networks to exploit spatial and temporal graph representations of human skeletons for 3D pose estimation. ...
arXiv:2301.07322v1
fatcat:vzmbarbql5hl3ksv4bronughry
PoseGraphNet++: Enriching 3D Human Pose with Orientation Estimation
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2023 pre-print arXiv:2308.11440v1 fatcat:sfahs7vh2jftnces4lkpflkncy
Existing skeleton-based 3D human pose estimation methods only predict joint positions. ...
We present PoseGraphNet++ (PGN++), a novel 2D-to-3D lifting Graph Convolution Network that predicts the complete human pose in 3D including joint positions and bone orientations. ...
METHOD We propose a Graph Convolution Network, named PoseG-raphNet++, for 3D human pose and orientation regression. ...
arXiv:2308.11440v2
fatcat:vaivpc2z7vbyxenn5i6w2hyvnu
HTNet: Human Topology Aware Network for 3D Human Pose Estimation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
3D human pose estimation errors would propagate along the human body topology and accumulate at the end joints of limbs. ...
Further considering the hierarchy of the human topology, joint-level and body-level dependencies are captured via graph convolutional networks and self-attentions, respectively. ...
Benefiting from the effective 2D HPE framework [3] , most recent works focus on the 2Dto-3D lifting pipeline [4] , which firstly estimates 2D poses from a single image and then lifts them to 3D keypoints ...
arXiv:2302.09790v1
fatcat:7qfy6537b5gotf5lsyjivpdy7q
A Single 2D Pose with Context is Worth Hundreds for 3D Human Pose Estimation
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Other Versions
The dominant paradigm in 3D human pose estimation that lifts a 2D pose sequence to 3D heavily relies on long-term temporal clues (i.e., using a daunting number of video frames) for improved accuracy, which ...
-- no finetuning on the 3D task is even needed. ...
Finally, a 1D convolution layer is adopted to gather temporal information, and a linear layer outputs the target 3D pose y ∈ R 1×(J×3) for the central video frame. ...
arXiv:2311.03312v2
fatcat:spoe5ojysfclfegayockvirgae
Deep Learning for 3D Human Pose Estimation and Mesh Recovery: A Survey
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
To the best of our knowledge, this survey is arguably the first to comprehensively cover deep learning methods for 3D human pose estimation, including both single-person and multi-person approaches, as ...
3D human pose estimation and mesh recovery have attracted widespread research interest in many areas, such as computer vision, autonomous driving, and robotics. ...
They designed a motion discriminator for adversarial training, utilizing datasets without any 3D labels, through a multi-scale spatio-temporal graph convolutional network. ...
arXiv:2402.18844v1
fatcat:hqfywkjouzbe3ifj36sdiidi2e
Graph Representation Learning and Its Applications: A Survey
2023
Sensors
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
In addition, we also discuss graph transformer models and Gaussian embedding models. ...
Over the decades, many models have been proposed for graph representation learning. ...
[47] Graph transformer https://github.com/mnschmit/graformer Graph-Bert [63] Graph transformer https://github.com/jwzhanggy/Graph-Bert EGT [30] Graph transformer https://github.com/shamim-hussain ...
doi:10.3390/s23084168
fatcat:pyhjz3plhbd67i3sj6xtyjrzzi