Article
Multidisciplinary Sciences
Tobias Baumgartner, Benjamin Paassen, Stefanie Klatt
Summary: Collecting large datasets for investigations into human locomotion is expensive and labor-intensive. Accurate methods for 3D human pose estimation in the wild could assist with collecting datasets for analyzing running kinematics from TV broadcast data. However, current state-of-the-art 3D human pose estimation methods are not yet accurate enough for kinematics research, as small differences in 3D angles play a significant role in biomechanical research.
SCIENTIFIC REPORTS
(2023)
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
Tong Li, Tianyun Dong
Summary: In this paper, a visual-inertial fusion approach is proposed to simplify the sensor calibration process in robot-assisted rehabilitation training, improving accuracy and suitability for patients who cannot perform poses accurately.
SENSORS AND ACTUATORS A-PHYSICAL
(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
Automation & Control Systems
Jia Hu, Shaoli Liu, Jianhua Liu, Zhenjie Wang, Wenxiong Zhang
Summary: In this article, a latent representation self-supervised pose network (LSPN) is proposed for accurate monocular pipe pose estimation. The network is trained with synthetic RGB data and uses a small number of labeled samples to establish the latent pose space, while a large number of structured unlabeled samples are used for self-supervised learning of latent pose representation. Experiments show that LSPN achieves excellent performance on real data and is robust to different environments, such as illumination changes and self-occlusion.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Information Systems
Yunjiao Zhou, He Huang, Shenghai Yuan, Han Zou, Lihua Xie, Jianfei Yang
Summary: This article proposes a WiFi-based Internet of Things-enabled human pose estimation scheme for metaverse avatar simulation. By mapping the channel state information of WiFi signals to human pose landmarks through self-attention, it effectively explores the spatial information of human pose. WiFi-based human poses, due to the ubiquity of WiFi and robustness to various illumination conditions, are suitable for instructing the movement of digital avatars in the metaverse, promoting avatar applications in smart homes.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Eduardo Souza dos Reis, Lucas Adams Seewald, Rodolfo Stoffel Antunes, Vinicius Facco Rodrigues, Rodrigo da Rosa Righi, Cristiano Andre da Costa, Luiz Gonzaga da Silveira Jr, Bjoern Eskofier, Andreas Maier, Tim Horz, Rebecca Fahrig
Summary: Multi-person pose estimation faces challenges such as person-to-person occlusion, truncated body parts, and double counting. Recent research has contributed successful methods to address these challenges. There is currently no up-to-date review on the latest advancements in tackling these challenges, and this study fills that gap.
PATTERN RECOGNITION
(2021)
Article
Chemistry, Analytical
Jiangying Zhao, Yongbiao Hu, Mingrui Tian
Summary: This study proposes a method for estimating the pose of excavator manipulators using computer vision technology, demonstrating its feasibility through experiments and error analysis while establishing a measurement system to simulate the pose estimation process.
Article
Engineering, Electrical & Electronic
Shaoxiang Guo, Eric Rigall, Yakun Ju, Junyu Dong
Summary: This article discusses the challenges of estimating 3D hand pose from a monocular RGB image and proposes a simple and efficient deep neural network to improve this task. By designing a feature chat block, the model is able to better handle the relationship between joint and skeleton features, resulting in improved accuracy and faster inference speed.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
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
Engineering, Aerospace
Zi Wang, Zhuo Zhang, Xiaoliang Sun, Zhang Li, Qifeng Yu
Summary: This article proposes an efficient monocular satellite pose estimation method by adopting transformer blocks and keypoint-set prediction network. The method achieves high-quality feature extraction through an effective satellite representation model and an improved backbone structure. Experimental results demonstrate its superior performance.
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS
(2022)
Article
Chemistry, Analytical
Xudong Zhang, Baigan Zhao, Jiannan Yao, Guoqing Wu
Summary: This paper presents a novel unsupervised learning framework for estimating scene depth and camera pose from video sequences. Multiple mask technologies and geometric consistency constraints are employed to mitigate the negative impacts of challenging scenes, such as dynamic objects and occluded regions. Experimental results on the KITTI dataset demonstrate that these strategies effectively enhance the model's performance, outperforming other unsupervised methods.
Article
Computer Science, Information Systems
Jinyoung Jun, Jae-Han Lee, Chul Lee, Chang-Su Kim
Summary: A novel monocular depth estimator is proposed to improve prediction accuracy on human regions by utilizing pose information. The algorithm consists of two networks, PoseNet and DepthNet, with a feature blending block and a joint training scheme to enhance depth estimation performance significantly.
Article
Automation & Control Systems
Changyu Zhao, Hirotaka Uchitomi, Taiki Ogata, Xianwen Ming, Yoshihiro Miyake
Summary: In this study, a method called OccCorrector was proposed to estimate 3D human poses robustly in occlusion cases involving a single camera and a few IMUs. The method includes a Sensor-Reshape module to fuse IMU information effectively and a new strategy of alternating the loss function to improve accuracy in predicting challenging poses.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Yantao Yu, Heng Li, Jiannong Cao, Xiaochun Luo
Summary: This article proposed a monocular-camera-based 3-D estimation method suitable for industrial working poses, utilizing a residual artificial neural network (RANN) with flexible complexity and weighted training loss. By establishing a 3-D pose data set containing diverse working poses in worksites, the network's performance in complex scenarios was evaluated. Compared to previous 3-D pose capture methods, the mean per joint position error was reduced by 31.42%, with a latency of 0.24 s. Hence, the proposed monocular-camera-based method shows great potential in industrial application scenarios.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Article
Computer Science, Theory & Methods
Jie Gui, Xiaofeng Cong, Yuan Cao, Wenqi Ren, Jun Zhang, Jing Zhang, Jiuxin Cao, Dacheng Tao
Summary: This article provides a comprehensive survey on supervised, semi-supervised, and unsupervised single image dehazing methods with the development of convolutional neural networks. It discusses the commonly used physical model, datasets, network modules, loss functions, and evaluation metrics. The main contributions of various dehazing algorithms are categorized and summarized, and experiments on baseline methods are conducted. Furthermore, the article points out unsolved issues and challenges for future research and provides a collection of useful dehazing materials.
ACM COMPUTING SURVEYS
(2023)
Article
Computer Science, Artificial Intelligence
Jinlong Fan, Jing Zhang, Dacheng Tao
Summary: This paper proposes a novel self-supervised image rectification method based on the idea that the rectified results of distorted images from different lenses should be the same. A new network architecture is designed with a shared encoder and multiple prediction heads, and a differentiable warping module is used to generate rectified and re-distorted images. The self-supervised learning scheme achieves comparable or better performance than the supervised baseline method and state-of-the-art methods.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Shengcong Chen, Changxing Ding, Minfeng Liu, Jun Cheng, Dacheng Tao
Summary: Nucleus segmentation is a challenging task due to the crowded distribution and blurry boundaries of nuclei. Recent approaches using polygons have achieved promising performance but lack contextual information. To address this, we propose a Context-aware Polygon Proposal Network (CPP-Net) that samples a point set for distance prediction, fuses predictions using a Confidence-based Weighting Module, and incorporates a novel Shape-Aware Perceptual (SAP) loss for constraining polygon shapes.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Chunhui Zhang, Guanjie Huang, Li Liu, Shan Huang, Yinan Yang, Xiang Wan, Shiming Ge, Dacheng Tao
Summary: This article proposes WebUAV-3M, the largest public UAV tracking benchmark to date, to facilitate the development and evaluation of deep UAV trackers. The benchmark dataset contains over 3.3 million frames across 4,500 videos and offers 223 highly diverse target categories. In addition, natural language specifications and audio descriptions are provided to take advantage of the complementary superiority of language and audio.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Automation & Control Systems
Mingjin Zhang, Qianqian Wu, Jing Zhang, Xinbo Gao, Jie Guo, Dacheng Tao
Summary: In this study, a novel super-resolution method based on fluid dynamics is proposed. The method abstracts the movement of pixels in the reconstruction process as the flow of fluid in fluid dynamics and utilizes a fluid micelle network to improve the detail reconstruction. Experimental results show that the proposed method outperforms other modern methods in terms of both objective metrics and visual quality.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Computer Science, Information Systems
Danuta Rutkowska, Piotr Duda, Jinde Cao, Leszek Rutkowski, Aleksander Byrski, Maciej Jaworski, Dacheng Tao
Summary: This paper presents a new incremental approach to mining data streams and focuses on tracking changes in the data stream. Probabilistic neural networks are used as basic models for this purpose. The paper introduces globally convergent stream data mining algorithms for regression, classification, and density estimation in a drifting environment. The algorithms are derived from Parzen kernel-based probabilistic neural networks and proven to have L-2 convergence. The paper provides illustrative examples for choosing the parameters and demonstrates the performance of the algorithms through simulations.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Zhe Chen, Jing Zhang, Yufei Xu, Dacheng Tao
Summary: Current object detectors commonly use a feature pyramid module for multi-level feature fusion to improve detection performance. However, they often require complex connections or iterative refinements, resulting in inefficient computation. To address this, we propose an efficient context modeling mechanism that enhances existing feature pyramids while reducing computational costs.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2023)
Article
Computer Science, Artificial Intelligence
Yadi Wang, Jun Wang, Dacheng Tao
Summary: Feature selection is a crucial technique for reducing dimensions in machine learning, pattern recognition, image processing, and data mining. Existing methods often suffer from sub-optimality due to their greedy nature. This paper introduces a supervised similarity measure-based holistic feature selection method, which outperforms several existing methods in terms of selecting informative features for classification.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Chen Gong, Yongliang Ding, Bo Han, Gang Niu, Jian Yang, Jane You, Dacheng Tao, Masashi Sugiyama
Summary: Label noise is a common issue in real-world scenarios, which misleads training algorithms and leads to degraded classification performance. This paper proposes a novel algorithm called Class-Wise Denoising (CWD) to tackle label noise by handling noisy labels in a class-wise manner, which improves classification performance under label noise.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Wei Xia, Quanxue Gao, Qianqian Wang, Xinbo Gao, Chris Ding, Dacheng Tao
Summary: In this article, we address the issues in existing graph-based multi-view clustering methods such as time burden and lack of consideration for both inter-view and intra-view similarity. We propose a variance-based de-correlation anchor selection strategy and a tensorized bipartite graph learning method to achieve multi-view clustering. The method exploits inter-view similarity by minimizing the tensor Schatten p-norm and exploits intra-view similarity using l(1,2)-norm minimization regularization and connectivity constraint. Experimental results demonstrate the superiority of our method over the state-of-the-art methods.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Automation & Control Systems
Mengya Han, Yibing Zhan, Baosheng Yu, Yong Luo, Han Hu, Bo Du, Yonggang Wen, Dacheng Tao
Summary: To address the issue of lacking robustness in concept learning for few-shot learning, we propose a novel metric-based method called region-adaptive concept aggregation network (RCA-Net). By designing a region-adaptive concept aggregator (RCA), we are able to better capture the conceptual information in different regions and obtain more accurate conceptual representation. By integrating the conceptual information from these regions in a weighted average manner, we can focus more on the concept-relevant information and improve the robustness of concept learning.
MACHINE INTELLIGENCE RESEARCH
(2023)
Article
Automation & Control Systems
Qi Zheng, Chao-Yue Wang, Dadong Wang, Da-Cheng Tao
Summary: Concept learning constructs visual representations connected to linguistic semantics for vision-language tasks. However, existing learners are vulnerable to attribute perturbations and out-of-distribution compositions. The paper proposes a framework for modeling semantic-aware visual subspaces and explores visual superordinates under linguistic hierarchy. Experimental results demonstrate the superiority of the proposed framework in diverse settings, improving overall answering accuracy by 7.5% for reasoning with perturbations and 15.6% for compositional generalization tests.
MACHINE INTELLIGENCE RESEARCH
(2023)
Article
Automation & Control Systems
Jing Jing, Ziyang Liu, Hao Guan, Wanlin Zhu, Zhe Zhang, Xia Meng, Jian Cheng, Yuesong Pan, Yong Jiang, Yilong Wang, Haijun Niu, Xingquan Zhao, Wei Wen, Jinxi Lin, Wei Li, Hao Li, Perminder S. S. Sachdev, Tao Liu, Zixiao Li, Dacheng Tao, Yongjun Wang
Summary: Ischemic strokes and transient ischemic attacks are leading causes of death worldwide. A deep learning-based risk prediction system has been developed to predict the probability of stroke recurrence or disability, which outperforms conventional risk scores and has the potential to optimize management in stroke patients.
ADVANCED INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Guanyu Gao, Yonggang Wen, Dacheng Tao
Summary: In this study, we propose a multiagent reinforcement learning approach with an attention mechanism to learn the optimal policies for microgrids without complex system modeling. Through evaluation using a simulation environment and real-world datasets, our method shows significant cost reduction for microgrids.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Zhen Wang, Liu Liu, Yiqun Duan, Dacheng Tao
Summary: This study proposes the task of few-shot streaming label learning (FSLL) and introduces a meta-learning framework (SIN) to adapt to this task. SIN leverages label semantic representation to regularize the output space and acquires labelwise meta-knowledge through meta-learning. Additionally, SIN incorporates a label decision module and meta-threshold loss function to determine the optimal confidence thresholds for each new label. Experimental results demonstrate that SIN outperforms the prior state-of-the-art methods on FSLL.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Rui Lv, Dingheng Wang, Jiangbin Zheng, Zhao-Xu Yang
Summary: In this paper, the authors investigate tensor decomposition for neural network compression. They analyze the convergence and precision of tensor mapping theory, validate the rationality of tensor mapping and its superiority over traditional tensor approximation based on the Lottery Ticket Hypothesis. They propose an efficient method called 3D-KCPNet to compress 3D convolutional neural networks using the Kronecker canonical polyadic (KCP) tensor decomposition. Experimental results show that 3D-KCPNet achieves higher accuracy compared to the original baseline model and the corresponding tensor approximation model.
Article
Computer Science, Artificial Intelligence
Xiangkun He, Zhongxu Hu, Haohan Yang, Chen Lv
Summary: In this paper, a novel constrained multi-objective reinforcement learning algorithm is proposed for personalized end-to-end robotic control with continuous actions. The approach trains a single model using constraint design and a comprehensive index to achieve optimal policies based on user-specified preferences.
Article
Computer Science, Artificial Intelligence
Zhijian Zhuo, Bilian Chen, Shenbao Yu, Langcai Cao
Summary: In this paper, a novel method called Expansion with Contraction Method for Overlapping Community Detection (ECOCD) is proposed, which utilizes non-negative matrix factorization to obtain disjoint communities and applies expansion and contraction processes to adjust the degree of overlap. ECOCD is applicable to various networks with different properties and achieves high-quality overlapping community detection.
Article
Computer Science, Artificial Intelligence
Yizhe Zhu, Chunhui Zhang, Jialin Gao, Xin Sun, Zihan Rui, Xi Zhou
Summary: In this work, the authors propose a Contrastive Spatio-Temporal Distilling (CSTD) approach to improve the detection of high-compressed deepfake videos. The approach leverages spatial-frequency cues and temporal-contrastive alignment to fully exploit spatiotemporal inconsistency information.
Review
Computer Science, Artificial Intelligence
Laijin Meng, Xinghao Jiang, Tanfeng Sun
Summary: This paper provides a review of coverless steganographic algorithms, including the development process, known contributions, and general issues in image and video algorithms. It also discusses the security of coverless steganography from theoretical analysis to actual investigation for the first time.
Article
Computer Science, Artificial Intelligence
Yajie Bao, Tianwei Xing, Xun Chen
Summary: Visual question answering requires processing multi-modal information and effective reasoning. Neural-symbolic learning is a promising method, but current approaches lack uncertainty handling and can only provide a single answer. To address this, we propose a confidence based neural-symbolic approach that evaluates NN inferences and conducts reasoning based on confidence.
Article
Computer Science, Artificial Intelligence
Anh H. Vo, Bao T. Nguyen
Summary: Interior style classification is an interesting problem with potential applications in both commercial and academic domains. This project proposes a method named ISC-DeIT, which combines data-efficient image transformer architectures and knowledge distillation, to address the interior style classification problem. Experimental results demonstrate a significant improvement in predictive accuracy compared to other state-of-the-art methods.
Article
Computer Science, Artificial Intelligence
Shashank Kotyan, Danilo Vasconcellos Vargas
Summary: This article introduces a novel augmentation technique called Dynamic Scanning Augmentation to improve the accuracy and robustness of Vision Transformer (ViT). The technique leverages dynamic input sequences to adaptively focus on different patches, resulting in significant changes in ViT's attention mechanism. Experimental results demonstrate that Dynamic Scanning Augmentation outperforms ViT in terms of both robustness to adversarial attacks and accuracy against natural images.
Article
Computer Science, Artificial Intelligence
Hiba Alqasir, Damien Muselet, Christophe Ducottet
Summary: The article proposes a solution to improve the learning process of a classification network by providing shape priors, reducing the need for annotated data. The solution is tested on cross-domain digit classification tasks and a video surveillance application.
Article
Computer Science, Artificial Intelligence
Dexiu Ma, Mei Liu, Mingsheng Shang
Summary: This paper proposes a method using neural dynamics solvers to solve infinity-norm optimization problems. Two improved solvers are constructed and their effectiveness and superiority are demonstrated through theoretical analysis and simulation experiments.
Article
Computer Science, Artificial Intelligence
Francesco Gregoretti, Giovanni Pezzulo, Domenico Maisto
Summary: Active Inference is a computational framework that uses probabilistic inference and variational free energy minimization to describe perception, planning, and action. cpp-AIF is a header-only C++ library that provides a powerful tool for implementing Active Inference for Partially Observable Markov Decision Processes through multi-core computing. It is cross-platform and improves performance, memory management, and usability compared to existing software.
Article
Computer Science, Artificial Intelligence
Zelin Ying, Dawei Cheng, Cen Chen, Xiang Li, Peng Zhu, Yifeng Luo, Yuqi Liang
Summary: This paper proposes a novel stock market trends prediction framework called SMART, which includes a self-supervised stock technical data sequence embedding model S3E. By training with multiple self-supervised auxiliary tasks, the model encodes stock technical data sequences into embeddings and uses the learned sequence embeddings for predicting stock market trends. Extensive experiments on China A-Shares market and NASDAQ market prove the high effectiveness of our model in stock market trends prediction, and its effectiveness is further validated in real-world applications in a leading financial service provider in China.
Article
Computer Science, Artificial Intelligence
Hao Li, Hao Jiang, Dongsheng Ye, Qiang Wang, Liang Du, Yuanyuan Zeng, Liu Yuan, Yingxue Wang, C. Chen
Summary: DHGAT1, a dynamic hyperbolic graph attention network, utilizes hyperbolic metric properties to embed dynamic graphs. It employs a spatiotemporal self-attention mechanism and weighted node representations, resulting in excellent performance in link prediction tasks.
Article
Computer Science, Artificial Intelligence
Jiehui Huang, Zhenchao Tang, Xuedong He, Jun Zhou, Defeng Zhou, Calvin Yu-Chian Chen
Summary: This study proposes a progressive learning multi-scale feature blending model for image deraining tasks. The model utilizes detail dilation and texture extraction to improve the restoration of rainy images. Experimental results show that the model achieves near state-of-the-art performance in rain removal tasks and exhibits better rain removal realism.
Article
Computer Science, Artificial Intelligence
Lizhi Liu, Zilin Gao, Yinhe Wang, Yongfu Li
Summary: This paper proposes a novel discrete-time interconnected model for depicting complex dynamical networks. The model consists of nodes and edges subsystems, which consider the dynamic characteristic of both nodes and edges. By designing control strategies and coupling modes, the stabilization and synchronization of the network are achieved. Simulation results demonstrate the effectiveness of the proposed methods.
