Filters








12 Hits in 2.9 sec

XNect: Real-time Multi-person 3D Human Pose Estimation with a Single RGB Camera [article]

Dushyant Mehta, Oleksandr Sotnychenko, Franziska Mueller, Weipeng Xu, Mohamed Elgharib, Pascal Fua, Hans-Peter Seidel, Helge Rhodin, Gerard Pons-Moll, Christian Theobalt
2019 arXiv   pre-print
We present a real-time approach for multi-person 3D motion capture at over 30 fps using a single RGB camera.  ...  This is a further key distinction from previous work that neither extracted global body positions nor joint angle results of a coherent skeleton in real time for multi-person scenes.  ...  CONCLUSION We present the first real-time approach for multi-person 3D motion capture using a single RGB camera.  ... 
arXiv:1907.00837v1 fatcat:7m5y6d7mtzfnhhgfq56jkm6nl4

Real-Time Hybrid Mapping of Populated Indoor Scenes using a Low-Cost Monocular UAV [article]

Stuart Golodetz, Madhu Vankadari, Aluna Everitt, Sangyun Shin, Andrew Markham, Niki Trigoni
2022 arXiv   pre-print
However, despite many recent works on both marker-based and markerless multi-UAV single-person motion capture, markerless single-camera multi-person 3D human pose estimation remains a much earlier-stage  ...  In this paper, we present what is thus, to our knowledge, the first system to perform simultaneous mapping and multi-person 3D human pose estimation from a monocular camera mounted on a single UAV.  ...  To our knowledge, no approach has yet deployed real-time markerless multi-person 3D human pose estimation for a monocular drone.  ... 
arXiv:2203.02453v1 fatcat:hs2ymgqzyfethl6eszh4iemhcm

Gravity-Aware Monocular 3D Human-Object Reconstruction [article]

Rishabh Dabral and Soshi Shimada and Arjun Jain and Christian Theobalt and Vladislav Golyanik
2021 arXiv   pre-print
., a new approach for joint markerless 3D human motion capture and object trajectory estimation from monocular RGB videos. We focus on scenes with objects partially observed during a free flight.  ...  We evaluate GraviCap on a new dataset with ground-truth annotations for persons and different objects undergoing free flights.  ...  Introduction Markerless 3D human motion capture from a single monocular RGB camera has many open challenges.  ... 
arXiv:2108.08844v1 fatcat:ve7om7j4kfcfvp76f2vs2gqjqu

Recovering 3D Human Mesh from Monocular Images: A Survey [article]

Yating Tian, Hongwen Zhang, Yebin Liu, Limin Wang
2022 arXiv   pre-print
Meanwhile, continuous efforts are devoted to improving the quality of 3D mesh labels for a wide range of datasets.  ...  With the same goal of obtaining well-aligned and physically plausible mesh results, two paradigms have been developed to overcome challenges in the 2D-to-3D lifting process: i) an optimization-based paradigm  ...  Currently, 1.5M frames with ground truth 3D skeletons from 65 sequences are available. [213] is a single-person 3D pose dataset collected in a multi-camera green screen studio.  ... 
arXiv:2203.01923v2 fatcat:vb6xa5wdsrhdxd2ebvg54qq2m4

Deep Learning Methods for 3D Human Pose Estimation under Different Supervision Paradigms: A Survey

Dejun Zhang, Yiqi Wu, Mingyue Guo, Yilin Chen
2021 Electronics  
The learning-based pose estimation is discussed from two categories: single-person and multi-person.  ...  Based on this literature survey, it can be concluded that each branch of 3D human pose estimation starts with fully-supervised methods, and there is still much room for multi-person pose estimation based  ...  [81] provide a real-time approach for multi-person 3D pose estimation, which is robust to difficult occlusion both by other people and objects.  ... 
doi:10.3390/electronics10182267 fatcat:ajnizu776ncpto3jvyh3zye2si

A review of 3D human pose estimation algorithms for markerless motion capture [article]

Yann Desmarais, Denis Mottet, Pierre Slangen, Philippe Montesinos
2021 arXiv   pre-print
Advances in convolutional networks triggered noticeable improvements in 2D pose estimation, leading modern 3D markerless motion capture techniques to an average error per joint of 20 mm.  ...  However, with the proliferation of methods, it is becoming increasingly difficult to make an informed choice.  ...  Single Person vs Multi-Person The strong assumptions that are still used for many 3D human pose estimation frameworks are related to camera or subject motion and acquisition protocol.  ... 
arXiv:2010.06449v3 fatcat:lo442jzoqvhlfedse7u7l6fjje

Reconstructing 3D Human Pose by Watching Humans in the Mirror [article]

Qi Fang, Qing Shuai, Junting Dong, Hujun Bao, Xiaowei Zhou
2021 arXiv   pre-print
In this paper, we introduce the new task of reconstructing 3D human pose from a single image in which we can see the person and the person's image through a mirror.  ...  The experiments demonstrate that, when trained on Mirrored-Human with our reconstructed 3D poses as pseudo ground-truth, the accuracy and generalizability of existing single-view 3D pose estimators can  ...  The effectiveness of our framework is validated on a new dataset for this new task with 3D pose ground-truth provided by a multi-view camera system.  ... 
arXiv:2104.00340v1 fatcat:q7bwzclmhrh7zfmneizglol4py

Orientation Keypoints for 6D Human Pose Estimation [article]

Martin Fisch, Ronald Clark
2021 arXiv   pre-print
Inspired by how motion-capture systems use a set of point markers to estimate full bone rotations, our method uses virtual markers to generate sufficient information to accurately infer rotations with  ...  In this paper we therefore introduce orientation keypoints, a novel approach for estimating the full position and rotation of skeletal joints, using only single-frame RGB images.  ...  XNect: [21] Catalin Ionescu, Dragos Papava, Vlad Olaru, and Cristian Smin- Real-time multi-person 3D motion capture with a single RGB chisescu.  ... 
arXiv:2009.04930v2 fatcat:bjahcsglunasxdahyitwouwanq

SmoothNet: A Plug-and-Play Network for Refining Human Poses in Videos [article]

Ailing Zeng, Lei Yang, Xuan Ju, Jiefeng Li, Jianyi Wang, Qiang Xu
2021 arXiv   pre-print
We conduct extensive experiments on twelve backbone networks with seven datasets across 2D and 3D pose estimation, body recovery, and downstream tasks.  ...  When analyzing human motion videos, the output jitters from existing pose estimators are highly-unbalanced.  ...  Xnect: Real-time multi- [31] Chang-Hung Lee, Cheng-Ru Lin, and Ming-Syan person 3d motion capture with a single rgb camera. Chen.  ... 
arXiv:2112.13715v1 fatcat:ldaslnqlzvb3few2qwezwdptou

Real-time 3D human body pose estimation from monocular RGB input [article]

Dushyant Mehta, Universität Des Saarlandes
2020
Conclusion This chapter presents the first real-time approaches for single-person and multi-person 3D motion capture using a single RGB camera.  ...  1.4 Overview of the general framework developed in this thesis for real-time 3D pose estimation from a single RGB camera in single person scenarios.  ...  A.1 Other Associated Data Captured The recording setup also consists of one camera with a fisheye lens and one depth camera.  ... 
doi:10.22028/d291-32536 fatcat:eenx4veqyrhlfbge5nyjikre6a

A Review of 3D Human Pose Estimation from 2D Images

Kristijan BARTOL, David BOJANIC, Tomislav PETKOVIC, Nicola D'APUZZO, Tomislav PRIBANIC
2020 Proceedings of 3DBODY.TECH 2020 - 11th International Conference and Exhibition on 3D Body Scanning and Processing Technologies, Online/Virtual, 17-18 November 2020   unpublished
A pose can be estimated from single or multiple frames, in a single (monocular) or multi-view (stereo) setup and for a single person or multiple people in the scene.  ...  In this work, we provide an overview of the classic and deep learning-based 3D pose estimation approaches.  ...  A convincing real-time performance was presented with 12 to 28 camera views.  ... 
doi:10.15221/20.29 fatcat:52ui5et4pbh7llrps2ojfmpxnu

A Tabletop for the Natural Inspection of Decorative Surfaces

Anton Kindsvater, Tom David Eibich, Martin Weier, André Hinkenjann
2021
Here, the complex interaction of light sources and the camera position must be controlled using software controls. As a result, only small parts of the data set can be properly inspected at a time.  ...  a potentially unnatural interaction with digital designs.  ...  Acknowledgements We would like to thank the people at CRUSE Imaging System for providing us with designs and PBR materials.  ... 
doi:10.2312/egve.20211329 fatcat:jfr4b7lvrvaljinll2xesayqwm