Article
Computer Science, Artificial Intelligence
Xiaoming Deng, Dexin Zuo, Yinda Zhang, Zhaopeng Cui, Jian Cheng, Ping Tan, Liang Chang, Marc Pollefeys, Sean Fanello, Hongan Wang
Summary: This paper investigates the impact of view-independent features on 3D hand pose estimation from a single depth image, and proposes a novel recurrent neural network model for 3D hand pose estimation. The model uses a cascaded 3D pose-guided alignment strategy for view-independent feature extraction and a recurrent hand pose module for modeling the dependencies among sequential aligned features. Experiments show that this method significantly improves the state-of-the-art accuracy on popular benchmarks with simple yet efficient alignment and network architectures.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Lele Wu, Zhenbo Yu, Yijiang Liu, Qingshan Liu
Summary: In this study, we propose a limb pose aware framework consisting of a kinematic constraint aware network and a trajectory aware temporal module to improve the 3D prediction accuracy of limb joint positions. By introducing relative bone angles and absolute bone angles as kinematic constraints, and incorporating a hierarchical Transformer network for trajectory estimation, we successfully alleviate the problem of errors accumulated along limbs and achieve promising results.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Computer Science, Artificial Intelligence
Yuanlu Xu, Wenguan Wang, Tengyu Liu, Xiaobai Liu, Jianwen Xie, Song-Chun Zhu
Summary: This paper proposes a pose grammar model for 3D human pose estimation from a monocular RGB image. The model leverages the estimated 2D pose as input and learns a mapping function to convert it into 3D pose. The model consists of a base network and bidirectional RNNs to capture pose-aligned features and incorporate knowledge about human body configuration. The research also improves model robustness through a data augmentation algorithm. Validation on 3D human pose benchmarks and cross-view evaluation demonstrates the effectiveness of the proposed method in handling these challenges.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Engineering, Electrical & Electronic
Tianlang Chen, Chen Fang, Xiaohui Shen, Yiheng Zhu, Zhili Chen, Jiebo Luo
Summary: This work proposes a new solution to 3D human pose estimation in videos. It decomposes the task into bone direction prediction and bone length prediction, drawing inspiration from human skeleton anatomy. The model performs high-accuracy bone length prediction utilizing global information and introduces a joint shift loss to bridge the training of the prediction networks. The model also utilizes an implicit attention mechanism to mitigate depth ambiguity in challenging poses, achieving superior performance compared to previous methods on benchmark datasets.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Engineering, Mechanical
G. Cepon, D. Ocepek, M. Kodric, M. Demsar, T. Bregar, M. Boltezar
Summary: This paper proposes the use of ArUco markers for impact-pose estimation in FRF acquisition. The approach relies on two dodecahedrons with markers, one mounted on the impact hammer and another at a known location on the structure. An experimental setup with an analog trigger is suggested, recording an image at the impact's exact time, from which the impact pose is estimated in the structure's coordinate system. Finally, a procedure to compensate for the location error is presented.
EXPERIMENTAL TECHNIQUES
(2023)
Article
Robotics
Mohammad Zubair, Sachin Kansal, Sudipto Mukherjee
Summary: This article discusses intervertebral motion in the craniovertebral junction (CVJ) region, emphasizing the need for performance evaluation of implants and the use of ArUco markers for accurate pose estimation.
Article
Chemistry, Analytical
Wenxia Bao, Zhongyu Ma, Dong Liang, Xianjun Yang, Tao Niu
Summary: In this paper, a self-supervised 3D pose estimation model called Pose ResNet is proposed, which uses 2D images as research objects and utilizes ResNet50 to extract features. The model introduces CBAM and WASP modules to refine selection of significant pixels and capture multi-scale contextual information. A deconvolution network is used to obtain the volume heat map, which is further processed to acquire joint coordinates. This model enables accurate estimation of 3D human pose from a single 2D image without the need for 3D ground truth labels.
Article
Computer Science, Artificial Intelligence
Shuangjun Liu, Naveen Sehgal, Sarah Ostadabbas
Summary: This paper presents an adapted human pose estimation approach to address the issue of domain shift in 3D human pose estimation. By using synthetic data and adaptive strategies, the proposed method achieves comparable performance with models trained on large-scale real datasets, and also provides a lightweight head to improve existing models.
APPLIED INTELLIGENCE
(2022)
Review
Engineering, Multidisciplinary
Jianchu Lin, Shuang Li, Hong Qin, Hongchang Wang, Ning Cui, Qian Jiang, Haifang Jian, Gongming Wang
Summary: This article summarizes the research progress and methods related to 3D human pose estimation, focusing on the estimation of monocular RGB images and videos. It discusses common problems such as occlusion and poor model generalization, and introduces various approaches, including multi-view and weakly supervised learning methods. The article also emphasizes the issue of insufficient training datasets and discusses emerging datasets and evaluation indicators. Overall, it provides a useful overview of the field and can guide researchers, while also highlighting areas for further research.
CMES-COMPUTER MODELING IN ENGINEERING & SCIENCES
(2023)
Article
Cell Biology
Pierre Karashchuk, Katie L. Rupp, Evyn S. Dickinson, Sarah Walling-Bell, Elischa Sanders, Eiman Azim, Bingni W. Brunton, John C. Tuthill
Summary: Anipose is an open-source toolkit for robust markerless 3D pose estimation, built on the 2D tracking method DeepLabCut. It enables users to expand existing experimental setups for accurate 3D tracking. Analysis of 3D leg kinematics tracked with Anipose reveals the key role of joint rotation in motor control of fly walking.
Article
Computer Science, Artificial Intelligence
Hai Ci, Xiaoxuan Ma, Chunyu Wang, Yizhou Wang
Summary: In this paper, an approach for estimating 3D human pose from monocular images is presented. By combining graph convolutional network (GCN) with locally connected network (LCN), the proposed approach achieves better performance on benchmark datasets and demonstrates strong generalization ability.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Xuefei Sun, Mohammed Jajere Adamu, Ruifeng Zhang, Xin Guan, Qiang Li
Summary: In order to accurately estimate human pose, we propose a novel method that directly incorporates pixel coordinates for pose estimation and uses coordinate attention to capture position- and structure-sensitive features. By using small-scale convolution kernels and smooth activation functions, we further reduce network parameters and computational cost. Through evaluation on the MPII Human Pose and COCO Keypoint Detection datasets, we demonstrate improved accuracy, highlighting the importance of directly incorporating coordinate location information in position-sensitive tasks.
Article
Chemistry, Multidisciplinary
Xiaowei Dai, Shuiwang Li, Qijun Zhao, Hongyu Yang
Summary: Animal pose estimation is important for analyzing behavior and health, but it often faces challenges in wild-animal images due to occlusions, backgrounds, and illumination conditions. This paper introduces a method that uses 3D prior constraints to improve the accuracy of animal pose estimation, by learning a 3D pose dictionary and optimizing representation coefficients.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Interdisciplinary Applications
Junjie Chen, Shuai Li, Donghai Liu, Weisheng Lu
Summary: This study proposes a mapping-free indoor localization method that transforms BIM renderings into photorealistic images using a CycleGAN model and estimates camera poses with a photogrammetry-based algorithm. Experimental results demonstrate that this method can achieve high-precision camera pose estimation.
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING
(2022)
Article
Computer Science, Hardware & Architecture
David Pascual-Hernandez, Nuria Oyaga de Frutos, Inmaculada Mora-Jimenez, Jose Maria Canas-Plaza
Summary: In this paper, the authors propose an end-to-end pipeline for real-time estimation of 3D human poses using a commercial RGBD sensor. The pipeline consists of two stages: 2D pose estimation using deep neural networks and 3D registration using a lightweight algorithm based on classic computer vision techniques. The authors compare different 2D pose estimation methods and validate their proposed method against a publicly available dataset. The results show high frame rates and low joint errors, and the method is agnostic to the 2D pose estimation model and can be upgraded or adapted for different articulated objects.
Article
Chemistry, Multidisciplinary
Ivan Virgala, L'ubica Mikova, Tatiana Kelemenova, Martin Varga, Robert Rakay, Marek Vagas, Jan Semjon, Rudolf Janos, Marek Sukop, Peter Marcinko, Peter Tuleja
Summary: The paper discusses a proposed concept for a biped robot that has vertical stabilization and minimizes sideways oscillation. It can be used as a mechatronic assistant or as a carrier for handling extensions.
APPLIED SCIENCES-BASEL
(2022)
Article
Chemistry, Analytical
Petr Oscadal, Tomas Spurny, Tomas Kot, Stefan Grushko, Jiri Suder, Dominik Heczko, Petr Novak, Zdenko Bobovsky
Summary: This work focuses on improving a camera system for sensing dynamic obstacles by implementing distributed data processing and the ability to change the number of cameras. Performance benchmark results show that the distributed system is faster in processing speed, framerate stability, and network data transmission efficiency compared to the centralized system.
Article
Engineering, Mechanical
Ivan Virgala, Michal Kelemen, Tatiana Kelemenova, Lubica Mikova, Erik Prada, Stefan Grushko, Martin Varga, Peter Jan Sincak, Tomas Merva, Zdenko Bobovsky
Summary: The article discusses the design of a robot with an unconventional kinematic structure, aiming to stabilize the robot base position with as few actuators as possible for sensor placement and handling superstructures.
MM SCIENCE JOURNAL
(2022)
Article
Engineering, Mechanical
Lukas Lestach, Michal Kelemen, Tatiana Kelemenova, Ivan Virgala, Lubica Mikova, Erik Prada, Darina Hroncova, Martin Varga, Peter Jan Sincak, Tomas Merva
Summary: The article discusses concepts for a two-legged walking robot, proposing three kinematics concepts and evaluating them to select the optimal one.
MM SCIENCE JOURNAL
(2022)
Article
Chemistry, Analytical
Petr Oscadal, Tomas Kot, Tomas Spurny, Jiri Suder, Michal Vocetka, Libor Dobes, Zdenko Bobovsky
Summary: Human-robot interaction is essential in practice, with an emphasis on workplace safety. Existing solutions control robots based on collision energy or re-planning trajectories. However, a sensor system is needed to detect obstacles in the shared workspace. Currently, engineers lack a procedure for deploying sensors optimally.
Article
Chemistry, Analytical
Ales Vysocky, Tomas Postulka, Jakub Chlebek, Tomas Kot, Jan Maslowski, Stefan Grushko
Summary: The article explores the possibilities of using hand gestures as a control interface for robotic systems in a collaborative workspace. The development of hand gesture control interfaces has become increasingly important in everyday life as well as professional contexts such as manufacturing processes. The results indicate that the gesture-based interface enables users to define a given path objectively faster than conventional methods.
Article
Engineering, Mechanical
Peter Tuleja, Rudolf Janos, Jan Semjon, Marek Sukop, Peter Marcinko
Summary: Technical solutions based on biological models, such as artificial muscles and compressed air actuators, were studied in this research. A test mechanism based on pneumatic muscles was constructed and measurements were made at different pressures. The results showed that pneumatic artificial muscles are suitable for imitating the human arm.
Article
Engineering, Electrical & Electronic
Jan Semjon, Rudolf Janos, Marek Sukop, Peter Tuleja, Peter Marcinko, Marek Nowakowski
Summary: This paper discusses the verification and comparison of a newly developed electric actuator. The actuator is designed for use in a walking robot for robotic football, placed in the robot's knee joint and upper part. It includes a harmonic precision gearbox and an absolute rotation sensor. The prototype consists of aluminum and 3D-printed parts. The selected parameters were verified according to ISO standard 9283, and compared with those of the standard actuator used in constructing robots for robotic football. The verification aims to improve performance parameters while ensuring accurate positioning by using a more accurate harmonic reducer and rotation sensor compared to the standard actuator.
Article
Engineering, Electrical & Electronic
Roman Mykhailyshyn, Frantisek Duchon, Ivan Virgala, Peter Jan Sincak, Ann Majewicz Fey
Summary: The article presents a numerical simulation model for airflow dynamics in the nozzle and radial gap of a Bernoulli gripping device. The optimal parameters of the device and the influence of nozzle geometry on pressure distribution and lifting force are determined. The C-Factor is used to determine the optimal gap height between the device and the manipulated object, enabling efficient operation of Bernoulli gripping devices in automated handling operations.
Article
Engineering, Mechanical
Martin Varga, Tatiana Kelemenova, Michal Kelemen, Ivan Virgala, Lubica Mikova, Erik Prada, Peter Marcinko, Marek Sukop, Dominik Novotny
Summary: This article presents a methodology for quickly identifying basic sensor parameters, which plays an important role in calibrating the liquid pressure sensor and integrating it into the selected application.
MM SCIENCE JOURNAL
(2023)
Article
Engineering, Mechanical
Tatiana Kelemenova, Michal Kelemen, Ivan Virgala, Lubica Mikova, Erik Prada, Martin Varga, Peter Marcinko, Marek Sukop, Dominik Novotny
Summary: The article discusses the methodology of verifying non-contact thermometers. A black body calibrator is used to evaluate the metrological properties of different temperature measurement devices, including industrial infrared thermometers, hand infrared thermometers, and infrared thermal cameras. The absolute measurement errors, standard deviations, and measurement uncertainties of these devices are monitored.
MM SCIENCE JOURNAL
(2023)
Article
Computer Science, Information Systems
Tomas Kot, Rostislav Wierbica, Petr Oscadal, Tomas Spurny, Zdenko Bobovsky
Summary: The paper presents a new version of the elastic band algorithm for path finding in the field of collaborative robotics. The algorithm dynamically modifies the position of control points on the path in reaction to obstacles in the workspace. It provides a smooth and length-optimal path while considering collisions more accurately than traditional methods.
Article
Computer Science, Information Systems
Daniel Huczala, Tomas Kot, Martin Pfurner, Vaclav Krys, Zdenko Bobovsky
Summary: This paper presents a set of algorithms for synthesizing the kinematic structures of serial manipulators and compares different kinematic representations. The results show that the choice of constraint design method significantly affects the success rate of optimization convergence, while the choice of representation has a lower impact on convergence but affects optimization time and manipulator length. The best results are obtained when multiple methodologies are used in combination, as demonstrated through an arbitrary manipulator designed in a collision environment.
Article
Engineering, Electrical & Electronic
Rudolf Janos, Marek Sukop, Jan Semjon, Peter Tuleja, Peter Marcinko, Martin Kocan, Maksym Grytsiv, Marek Vagas, L'ubica Mikova, Tatiana Kelemenova
Summary: Robotic football with humanoid robots is a multidisciplinary field. This study aims to determine a walking pattern for a humanoid robot with an impact on its dynamic stability and behavior. The design of the proposed technical concept depends on stability management mechanism, walking speed, and chosen stability approaches. The complexity of the walking principle and control of the robot's movement limit the versatility and adaptability of the humanoid robot.