Search

Your search keyword '"WANG, ZHENHUA"' showing total 269 results
Start Over Author "WANG, ZHENHUA" Publisher ieee
269 results on '"WANG, ZHENHUA"'

9. ISDA: Position-Aware Instance Segmentation with Deformable Attention

Author

Ying, Kaining, Wang, Zhenhua, Bai, Cong, and Zhou, Pengfei

Subjects
FOS: Computer and information sciences, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition
Abstract

Most instance segmentation models are not end-to-end trainable due to either the incorporation of proposal estimation (RPN) as a pre-processing or non-maximum suppression (NMS) as a post-processing. Here we propose a novel end-to-end instance segmentation method termed ISDA. It reshapes the task into predicting a set of object masks, which are generated via traditional convolution operation with learned position-aware kernels and features of objects. Such kernels and features are learned by leveraging a deformable attention network with multi-scale representation. Thanks to the introduced set-prediction mechanism, the proposed method is NMS-free. Empirically, ISDA outperforms Mask R-CNN (the strong baseline) by 2.6 points on MS-COCO, and achieves leading performance compared with recent models. Code will be available soon., Comment: Accepted to ICASSP 2022

Published
2022

14. Object Clustering With Dirichlet Process Mixture Model for Data Association in Monocular SLAM.

Author

Wei, Songlin, Chen, Guodong, Chi, Wenzheng, Wang, Zhenhua, and Sun, Lining

Subjects
MONOCULARS, OBJECT recognition (Computer vision), OFFICE environment, DATA modeling, CAMERAS
Abstract

Semantic simultaneous localization and mapping (SLAM) with a monocular camera is particularly attractive because of the deployment simplicity and economic availability. Data association problem which assigns unique identities for objects shown in multiple frames plays a fundamental role in semantic SLAM. Previous prevalent methods which mainly focused on associating geometric KeyPoints are no longer suitable. Some naive methods that rely on object distance or 2-D/3-D Intersection over Union are also vulnerable when occlusions happen. In this article, we propose a novel data association method for cuboid landmarks based on Dirichlet process mixture model. By jointly considering object class, position, and size, our method can perform data association robustly. We evaluated our method in simulated datasets, public benchmark KITTI, and on a real robot in an office environment. Experimental results show that our method not only associates cuboids robustly but also achieves SOTA pose estimation accuracy in monocular SLAMs. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

27. Graph Attention Tracking

Author

Guo, Dongyan, primary, Shao, Yanyan, additional, Cui, Ying, additional, Wang, Zhenhua, additional, Zhang, Liyan, additional, and Shen, Chunhua, additional

Published
2021
Full Text
View/download PDF

29. Event-Based Fault Detection for Unmanned Surface Vehicles Subject to Denial-of-Service Attacks.

Author

Fei, Zhongyang, Wang, Xudong, and Wang, Zhenhua

Subjects
DENIAL of service attacks, AUTONOMOUS vehicles, REMOTELY piloted vehicles, EXPONENTIAL stability, TELECOMMUNICATION systems, ENERGY consumption
Abstract

In this work, the problem of event-based fault detection is addressed for unmanned surface vehicles (USVs) under denial-of-service (DoS) jamming attacks. A fault detection filter (FDF) is deployed to generate residual for USV systems with external disturbance and system faults. A resilient event-triggered mechanism is implemented to reduce the bandwidth occupation of communication network and energy consumption of USV system as well as mitigating the influence of DoS attacks. Based on the established framework, a switched residual system is constructed and a criterion is deduced to ensure the exponential stability and weighted $H_{\infty }$ performance of residual system via piecewise Lyapunov functional approach. The FDF and resilient event-triggered mechanism are co-designed. Simulation results are provided to testify the efficient performance of the devised resilient event-triggered fault detection method. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

30. Interval-Observer-Based Fault Detection and Isolation Design for T-S Fuzzy System Based on Zonotope Analysis.

Author

Zhu, Fanglai, Tang, Yuyan, and Wang, Zhenhua

Subjects
LINEAR matrix inequalities, KALMAN filtering, MATRIX inequalities, DYNAMICAL systems
Abstract

This article deals with the fault detection and isolation problems for a class of uncertain discrete-time Takagi-Sugeno (T-S) fuzzy system based on the combination of the H-infinity observer and the zonotope method (ZM). For fault detection (FD) purpose, first, a Luenberger-like H-infinity observer, which is robust to disturbance in a sense of H-infinity performance index is designed under the assumption of the feasibility of an linear matrix inequality. Second, the ZM is applied to the H-infinity observer error dynamic system such that the interval state estimation can be calculated iteratively if the system suffers from neither actuator nor sensor faults. Third, a residual is constructed and its interval estimation is also given, and furthermore, based on the residual interval estimation, an FD scheme is developed. After this, we discuss the fault isolation issue in the similar way to alarm the appearance of the exact type fault: actuator or sensor fault. It is the ZM is applied onto the Luenberger-like observer, the fault detection and isolation performances are improved greatly. Finally, a numerical simulation example is given and some comparisons are also made to the existing results to verify the effectiveness and to show the advantages of proposed method. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

34. Temperature Evolution Properties of Overhead Line Conductor Exposed to Large-scale Jet Fires Induced by High-pressure Natural Gas Leakage

Author

Jiang Juncheng, Fei You, Yun Zhang, Jixiang Xu, Huangfu Wenhao, Kai Shui, Wang Zhenhua, and Zonglin Fu

Subjects
Electric power system, Electric power transmission, Materials science, Transmission line, Thermal, Mechanics, Electrical conductor, Overhead line, Conductor, Leakage (electronics)
Abstract

A unified model was developed from leaked high-pressure natural gas (up to 6.55 MPa) and a procedure outlined in IEEE Standard 738–2012 for transmission line conductor temperature. A real-scale case was presented to show the thermal effects of a large-scale jet fire on the nearby transmission line by setting their distances from 200 to 500 m. The predicted total flame height and damage radius correspond to the reported results with appropriate correction of input parameters. Transmission lines with higher load currents and shorter distances to the jet fire induce higher temperature rises in line conductors. Reducing the current load of transmission line needs to integrate both factors of thermal failure of transmission line and economy of power system.

Published
2019

40. Fault Detection for Lipschitz Nonlinear Systems With Restricted Frequency-Domain Specifications.

Author

Li, Jitao, Wang, Zhenhua, Ahn, Choon Ki, and Shen, Yi

Subjects
*NONLINEAR systems, *SYMMETRIC matrices, *FAULT diagnosis, *FREQUENCY-domain analysis
Abstract

This article deals with the problem of fault detection for discrete-time Lipschitz nonlinear systems subject to a class of restricted frequency-domain specifications. We present a novel observer structure with more design parameters, which can be applied to enhance the observer performance. The performances of fault sensitivity and disturbance robustness are characterized using finite-frequency $H_{-}$ and $H_{\infty }$ indices, respectively. Less restrictive design conditions are obtained based on a reformulated Lipschitz property. Moreover, to detect faults timely, a novel dynamic threshold is synthesized based on zonotopic set-membership techniques. Simulation examples are conducted to demonstrate the viability and validity of the presented method. [ABSTRACT FROM AUTHOR]

Published
2021
Full Text
View/download PDF

41. Self-Rectifying Al 2 O 3 /TaO x Memristor With Gradual Operation at Low Current by Interfacial Layer.

Author

Zhao, Xuanyu, Zhang, Kailiang, Hu, Kai, Zhang, Yujian, Zhou, Qiaozhen, Wang, Zhenhua, She, Yu, Zhang, Zhenzhong, and Wang, Fang

Subjects
ALUMINUM oxide, STRAY currents, MEMRISTORS
Abstract

Sneak path current phenomenon in a memristor array based on cross-point structure could lead to crosstalk, which limits its application in high-density large-scale arrays. Self-rectifying devices are the best choice for suppressing leakage currents without compromising the original performance of the device. It eliminates the need for selectors and effectively reduces process costs. In this work, a self-rectifying memristor based on Au/TaOx/Al2O3/TiN is proposed. Nonlinearity of ~236.76 with a sneak current below 1 nA with a tunable conductance under $1 \boldsymbol {\mu } \text{A}$ compliance current by introducing a rectifying switching mode after constructing the TaOx/Al2O3 different stoichiometric stack. Detailed analysis suggested that the interface barrier and the vacancy-related defects in Al2O3 are responsible for the special behavior. The effective array scale up to ~400 by readout margin calculation further confirms that the memristor can restrain the sneak current. Our self-rectifying memristor simplifies the device design of subsequent arrays, paving the way for addressing applications in highly compact scenes. [ABSTRACT FROM AUTHOR]

Published
2021
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results