Article
Engineering, Electrical & Electronic
Xianglong Zeng, Chaoyang Wu, Wen-Bin Ye
Summary: In recent years, radar-based dynamic hand gesture recognition systems have been widely used, researchers typically use machine learning algorithms, especially deep learning algorithms for high precision gesture recognition. However, most deep learning models only accept predefined gestures as input, not allowing user-defined gestures. This study proposes a neural network model trained with meta-learning method to handle few-shot classification tasks and enable user-definable DHGR.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
Zhaocheng Yang, Xinbo Zheng
Summary: The paper introduces a touchless hand gesture recognition method using radar sensor, including range-Doppler-angle trajectories extraction and a reused LSTM network. This method utilizes a 77GHz radar and a gesture desktop to improve recognition accuracy.
IEEE SENSORS JOURNAL
(2021)
Article
Geochemistry & Geophysics
Yong Wang, Yuhong Shu, Xiuqian Jia, Mu Zhou, Liangbo Xie, Lei Guo
Summary: This letter presents a method using FMCW radar for short-range hand gesture sensing and recognition. The range, Doppler, and angle parameters of hand gestures are measured using FFT and MUSIC algorithms. The proposed CMFF-HGR method extracts features and achieves hand gesture recognition.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Xiangyu Shen, Haifeng Zheng, Xinxin Feng, Jinsong Hu
Summary: Radar-based hand gesture recognition has gained much attention in the field of human-computer interaction due to its high accuracy and independence from lighting conditions. This paper proposes a meta-learning approach for the few-shot learning problem in frequency modulated continuous wave radar-based hand gesture recognition. Experimental results show that the proposed meta-learning model significantly improves recognition accuracy compared to state-of-the-art methods.
IEEE SENSORS JOURNAL
(2022)
Article
Engineering, Electrical & Electronic
Biao Jin, Yu Peng, Xiaofei Kuang, Zhenkai Zhang, Zhuxian Lian, Biao Wang
Summary: A robust hand gesture recognition method based on self-attention time-series neural networks is proposed in this paper to sense the subtle movement of the hand using millimeter-wave radar. The method achieves a high recognition rate in dynamic interference scenarios.
IEEE SENSORS JOURNAL
(2022)
Article
Engineering, Electrical & Electronic
Aloysius Adya Pramudita, Lukas, Edwar
Summary: This paper proposes an array radar configuration for contactless hand gesture sensor to achieve more accurate hand gesture features with 96.6% accuracy and simple data processing.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
Haoming Liu, Zhenyu Liu
Summary: This paper proposes a multimodal dynamic hand gesture recognition method based on a two-branch fusion deformable network with Gram matching. It effectively improves the adaptability of the classifier to complex environments and exhibits satisfactory robustness to multiple subjects.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Information Systems
Josiah W. Smith, Shiva Thiagarajan, Richard Willis, Yiorgos Makris, Murat Torlak
Summary: This paper investigates new techniques for data collection and training to improve the accuracy of non-moving hand gestures classification using deep CNN and mmWave radars. The proposed methods significantly enhance the classification rates for static hand gestures.
Article
Computer Science, Artificial Intelligence
Derek W. Orbaugh Antillon, Christopher R. Walker, Samuel Rosset, Iain A. Anderson
Summary: A smart dive glove has been developed to recognize 13 hand gestures used in diving communication. Different classification algorithms were trained and evaluated, showing good performance in a dry environment but decreasing underwater, especially when divers need to focus on other tasks.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Guan Yuan, Xiao Liu, Qiuyan Yan, Shaojie Qiao, Zhixiao Wang, Li Yuan
Summary: This study introduces a novel data glove and proposes an improved deep feature fusion network, achieving good results in gesture recognition, especially in recognizing American Sign Language and Chinese Sign Language.
IEEE SENSORS JOURNAL
(2021)
Article
Computer Science, Information Systems
Lesong Jia, Xiaozhou Zhou, Chengqi Xue
Summary: This paper presents a novel non-trajectory-based gesture recognition method (NT-GRM) based on hand skeleton information and a hidden Markov model (HMM). It can accurately and quickly recognize static and dynamic gestures with high recognition accuracy and speed.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Anil Osman Tur, Hacer Yalim Keles
Summary: In this study, a framework is proposed to combine deep features with HMM models for isolated sign recognition. Experimental results demonstrate that HMM can classify isolated signs with high accuracy on the Montalbano dataset.
MULTIMEDIA TOOLS AND APPLICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Ali Safa, Andre Bourdoux, Ilja Ocket, Francky Catthoor, Georges G. E. Gielen
Summary: Radar processing using spiking neural networks (SNNs) is a solution for ultralow-power wireless human-computer interactions. SNNs are significantly more energy-efficient compared to traditional deep learning methods, with successful application in radar gesture recognition, and providing evaluation code for future research.
IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Yi Zhang, Shuqin Dong, Chengkai Zhu, Marcel Balle, Bin Zhang, Lixin Ran
Summary: This research presents a non-contact solution for hand gesture recognition on smart devices using radar sensors, which classify gestures through machine learning algorithms. Experimental results validate the robustness and accuracy of this method.
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES
(2021)
Article
Engineering, Electrical & Electronic
Yingxiang Sun, Haoqiu Xiong, Danny Kai Pin Tan, Tony Xiao Han, Rui Du, Xun Yang, Terry Tao Ye
Summary: This study introduces a dual-frequency continuous wave radar that can achieve both localization and activity/gesture recognition simultaneously. Features of different movements are classified by the lightweight network AGRNet, and data corresponding to walking are used for moving target localization by comparing phase differences in the Doppler domain between dual frequencies. Additionally, a segmentation method is proposed to effectively extract individual time-periods corresponding to different motions from continuous signals. Experimental results demonstrate classification accuracy over 91% and centimeter-level localization accuracy.
IEEE SENSORS JOURNAL
(2021)
Article
Computer Science, Artificial Intelligence
Qun Guo, Jing Li, Fengdao Zhou, Gang Li, Jun Lin
Summary: In this paper, a novel fault diagnosis framework called Multiscale-AAM-OTCN is proposed to solve the known and unknown fault diagnosis problems of modular multilevel converters (MMCs) by outputting current signals. The framework incorporates batch normalization, layer normalization, and a multiscale coordinate residual attention mechanism to improve the convergence and generalization ability of the model for different tasks. Experimental results show the effectiveness of the proposed framework for both known and unknown fault diagnoses.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Information Systems
Shufan Yang, Julien Le Kernec, Olivier Romain, Francesco Fioranelli, Pierre Cadart, Jeremy Fix, Chenfang Ren, Giovanni Manfredi, Thierry Letertre, Israel David Hinostroza Saenz, Jifa Zhang, Huaiyuan Liang, Xiangrong Wang, Gang Li, Zhaoxi Chen, Kang Liu, Xiaolong Chen, Jiefang Li, Xing Wu, Yichang Chen, Tian Jin
Summary: Radar technology is valuable for monitoring human activities and can help elderly people live independently at home for longer. However, there are challenges in developing effective algorithms for radar-based human activity classification and validating them. To foster exploration and evaluation of different algorithms, a dataset released in 2019 was used to benchmark various classification approaches. This paper provides an overview and evaluation of the methods used in the inaugural Radar Challenge.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Chemistry, Multidisciplinary
Tianyu Zhang, Chunran Zhang, Donghui Long, Yanzhang Wang, Haigen Zhou, Shilong Wang, Gang Li, Haoran Li, Fengdao Zhou, Chuandong Jiang
Summary: This study conducted a qualitative survey on the ground-to-air and airborne electromagnetic detection methods used at the Huola Mountain Tunnel site, and analyzed the apparent resistivity profile and geological mapping data. The study identified major stratigraphic boundaries, fault fracture zones, rock fragmentation, weakness, karst development, and water content based on the apparent resistivity profile. The results provide guidance for the construction of the Huola Mountain Tunnel.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Electrical & Electronic
Guchong Li, Gang Li, You He
Summary: This article proposes a solution for distributed multiple resolvable group targets tracking (DMRGTT) by utilizing the labeled multi-Bernoulli (LMB) filter and distributed fusion rules including generalized covariance intersection (GCI) and weighted arithmetic average (WAA). The hypergraph matching (HM) theory is adopted to address the label matching (LM) problem caused by the proximity of targets' positions within the same group and the label mismatching problem. The experiment results demonstrate the effectiveness and robustness of the proposed fusion approaches.
IEEE SENSORS JOURNAL
(2023)
Article
Chemistry, Analytical
Yuan Zhang, Yixue Qiao, Gang Li, Wei Li, Qing Tian
Summary: Automotive radar aims to achieve low cost and high performance, especially by improving angular resolution without increasing the number of radar channels. This paper proposes a random time division multiplexing MIMO radar, which combines non-uniform linear array (NULA) and random time division transmission mechanism. It utilizes tensor completion technology to recover a sparse three-order receiving tensor and successfully completes range, velocity, and angle measurements. Simulations confirm the effectiveness of this method.
Article
Nanoscience & Nanotechnology
Han Liu, Xinyu Cao, Fengdao Zhou, Gang Li
Summary: In order to ensure the safe and reliable operation of Li-ion battery energy storage systems, this study proposes an online fusion estimation method based on BP-GA to estimate the state of charge (SoC) and state of health (SoH) of Li-ion batteries. The method analyzes the effective features of SoC and SoH, and uses a neural network to estimate them. The results show that the BP-GA method achieves higher accuracy and efficiency compared to the conventional BPNN method.
Article
Chemistry, Analytical
Jigen Xia, Ronghua Peng, Zhiqiang Li, Junyi Li, Yizhuo He, Gang Li
Summary: The development of underground artificial cavities is crucial for the exploitation of urban spatial resources. Geophysical techniques have been widely used in the construction, management, and maintenance of underground artificial cavities. This study introduces two identification methods for underground artificial cavities, including apparent resistivity imaging and a fast recognition approach based on the Bayesian convolutional neural network (BCNN).
Article
Energy & Fuels
Han Liu, Tao Han, Shangshuai Hao, Gang Li
Summary: This study proposes a one-dimensional voltage-correlated convolutional neural network (1DVCNN) method to detect and localize internal short circuit (ISC) faults in energy storage systems. The voltage signal of the LiFePO4 battery is converted into a Pearson Correlation Coefficient (PCC) through correlation analysis and used as features for fault diagnosis in a one-dimensional convolutional neural network (1DCNN). The results demonstrate that the proposed method can detect ISC faults in energy storage systems in a more timely and effective manner compared to other studies.
Article
Automation & Control Systems
Gang Li, Luge Wang, Chaochen Li
Summary: This paper proposes a control method for LED driver with light intensity feedback, and employs a fuzzy PID control strategy to improve steady-state accuracy and dynamic performance. The feasibility of this control method is demonstrated through the mathematical model and experimental verification.
INTERNATIONAL JOURNAL OF DYNAMICS AND CONTROL
(2023)
Article
Geochemistry & Geophysics
Chuan Qin, Xueqian Wang, Gang Li, You He
Summary: In this article, a novel semi-soft label-guided network based on self-distillation (S2LSDNet) is proposed for ship detection in SAR images. The proposed method extracts semi-soft label information using an efficient self-distillation training strategy, enhancing detection accuracy in complex inshore scenarios. Additionally, an angle-related and balanced IoU loss is developed to improve ship positioning performance. Experimental results show that the proposed method outperforms existing state-of-the-art approaches, particularly in inshore scenes.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Zhihao Wang, Xueqian Wang, Wei Wu, Gang Li
Summary: In this article, a spatiotemporal fusion CD (STFCD) algorithm is proposed to improve flood change detection performance. The algorithm utilizes the spatial dependence and temporal interaction of multisource heterogeneous satellite image time series (SITS).
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Tongyang Ren, Tao Han, Qun Guo, Gang Li
Summary: To improve the accuracy and confidence of power converter fault diagnosis, it is necessary to understand the change and decision mechanism inside the deep model. This study utilizes a temporal convolutional network (TCN) to visualize the diagnostic process, analyzes the effect of hyperparameters on generalizability, and interprets the concern area of the model for fault decision using gradient-weighted class activation mapping (Grad-CAM). The visualization results help improve the understanding of neural networks and support the design of more generalizable models.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Geochemistry & Geophysics
Jiannan Chen, Yu Liu, Xueqian Wang, Yiming Zhang, Zhizhuo Jiang, Gang Li, Bolun Zheng, Jiyong Zhang, You He
Summary: Accurate estimation of ship sizes is crucial for ship classification in SAR images. Existing DNNs-based methods for SAR ship size estimation (SSE) have limited capability in accurately modeling the relationships of features, leading to degraded performance. We propose an improved method based on capsule network (Caps-SSENet) that captures feature relationships using capsules and dynamic routing. Experimental results show that our method outperforms existing state-of-the-art methods in reducing ship size estimation error in SAR images.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Weiming Li, Lihui Xue, Xueqian Wang, Gang Li
Summary: Change detection in optical remote sensing images has been improved by deep convolutional neural networks (CNNs) and transformer modules. However, the existing simple cascade of CNNs and transformers shows limited performance on small changed areas. To address this issue, we propose a new ConvTransNet with a multiscale framework to better exploit global-local information in optical remote sensing images.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Dehua Wang, Gang Li, Zhichun Zhao, Jianwen Wang, Shuai Ding, Kunpeng Wang, Meiya Duan
Summary: The uneven and multicluster distribution of micro-motion feature vectors of space targets limits the performance of traditional radar anomaly detection methods. To address this issue, a novel detection method that utilizes local density peaks (LDPs) and micro-motion features is proposed in this letter. Discriminative micro-motion features are extracted from radar echoes to construct a 2-D feature space, and the abnormal motion detector is obtained by classifying feature vectors into different clusters based on LDPs and minimum spanning tree clustering (LDP-MST), and determining decision thresholds for each cluster using LDPs, neighbors, and preset false alarm rates. The electromagnetic simulation experiment results demonstrate significantly higher detection rates for the proposed method compared to six state-of-the-art methods, with an improvement ranging from 2.49% to 49.45% when the false alarm rate is 5%.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)