检索结果-内蒙古大学图书馆

BiN-Flow: Bidirectional Normalizing Flow for Robust Image Dehazing

IEEE TRANSACTIONS ON IMAGE PROCESSING 2022年 31卷 6635-6648页

作者： Wu, Yiqiang Tao, Dapeng Zhan, Yibing Zhang, Chenyang Yunnan Univ Sch Informat Sci & Engn FIST Lab Kunming 650500 Yunnan Peoples R China Yunnan Key Lab Intelligent Syst & Comp Kunming 650500 Yunnan Peoples R China Yunnan United Vis Technol Co Ltd Kunming 650500 Yunnan Peoples R China JD Explore Acad Beijing 100000 Peoples R China Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China

Image dehazing aims to remove haze in images to improve their image quality. However, most image dehazing methods heavily depend on strict prior knowledge and paired training strategy, which would hinder generalization and performance when dealing with unseen scenes. In this paper, to address the above problem, we propose Bidirectional Normalizing Flow (BiN-Flow), which exploits no prior knowledge and constructs a neural network through weakly-paired training with better generalization for image dehazing. Specifically, BiN-Flow designs 1) Feature Frequency Decoupling (FFD) for mining the various texture details through multi-scale residual blocks and 2) Bidirectional Propagation Flow (BPF) for exploiting the one-to-many relationships between hazy and haze-free images using a sequence of invertible Flow. In addition, BiN-Flow constructs a reference mechanism (RM) that uses a small number of paired hazy and haze-free images and a large number of haze-free reference images for weakly-paired training. Essentially, the mutual relationships between hazy and haze-free images could be effectively learned to further improve the generalization and performance for image dehazing. We conduct extensive experiments on five commonly-used datasets to validate the BiN-Flow. The experimental results that BiN-Flow outperforms all state-of-the-art competitors demonstrate the capability and generalization of our BiN-Flow. Besides, our BiN-Flow could produce diverse dehazing images for the same image by considering restoration diversity.

关键词： Training Image restoration Image color analysis Band-pass filters Convolutional neural networks Atmospheric modeling Noise measurement Image dehazing bidirectional normalizing flow weakly-paired training and restoration diversity

来源：评论

学校读者我要写书评

暂无评论

Local and Global Perception Generative Adversarial Network for Facial Expression Synthesis

引用

IEEE TRANSACTIONS ON CIRCUITS AND systems FOR VIDEO TECHNOLOGY 2022年第3期32卷 1443-1452页

作者： Xia, Yifan Zheng, Wenbo Wang, Yiming Yu, Hui Dong, Junyu Wang, Fei-Yue Univ Portsmouth Sch Creat Technol Portsmouth PO1 2DJ Hants England Xi An Jiao Tong Univ Sch Software Engn Xian 710049 Peoples R China Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Ocean Univ China Dept Informat Sci & Technol Qingdao 266100 Peoples R China Macau Univ Sci & Technol Inst Syst Engn Macau 999078 Peoples R China Qingdao Acad Intelligent Ind Qingdao 266109 Peoples R China Chinese Acad Sci State Key Lab Management & Control Complex Syst Inst Automat Beijing 100190 Peoples R China

Facial expression synthesis has gained increasing attention with the development of Generative Adversarial Networks (GANs). However, it is still very challenging to generate high-quality facial expressions since the overlapping and blur commonly appear in the generated facial images especially in the regions with rich facial features such as eye and mouth. Generally, existing methods mainly consider the face as a whole in facial expression synthesis without paying specific attention to the characteristics of facial expressions. In fact, according to the physiological and psychological research, the differences of facial expressions often appear in crucial regions such as eye and mouth. Motivated by this observation, a novel end-to-end facial expression synthesis method called Local and Global Perception Generative Adversarial Network (LGP-GAN) with a two-stage cascaded structure is proposed in this paper which is designed to extract and synthesize the details of the crucial facial regions. LGP-GAN can combine the generated results from the global network and local network into the corresponding facial expressions. In Stage I, LGP-GAN utilizes local networks to capture the local texture details of the crucial facial regions and generate local facial regions, which fully explores crucial facial region domain information in facial expressions. And then LGP-GAN uses a global network to learn the whole facial information in Stage II to generate the generate final facial expressions building upon local generated results from Stage I. We conduct qualitative and quantitative experiments on the commonly used public database to verify the effectiveness of the proposed method. Experimental results show the superiority of the proposed method over the state-of-the-art methods.

关键词： Task analysis Face recognition Mouth Generative adversarial networks Facial features Generators Gallium nitride Facial expression synthesis generative adversarial networks facial expression recognition local facial region facial mask

来源：评论

学校读者我要写书评

暂无评论

Pedestrian Trajectory Prediction by Modeling the Interactions using Social LSTM Extensions

Pedestrian Trajectory Prediction by Modeling the Interaction...

引用

2022 Chinese Automation Congress, CAC 2022

作者： Si, Zeze Zhao, Hongxia Tang, Haina Zhu, Fenghua Lv, Yisheng Lu, Hao Institute of Autamation Chinese Academy of Sciences State Key Laboratory for Management and Control of Complex Systems Beijing100190 China University of Chinese Academy of Sciences School of Artificial Intelligence Beijing100049 China Qingdao Academy of Intelligent Industries Qingdao266000 China

ISBN: (纸本)9781665465335

In many driving situations, human mobility is an important topic in trajectory prediction. Considering the pedestrian trajectory as a sequence generative task, a prediction algorithm based on Social Long Short-Term Memory (Social LSTM) is implemented. In order to simulate the social interaction between pedestrians, Social Pooling (S-Pooling) is used to aggregate the hidden state of pedestrians, while the attention mechanism is utilized to aggregate information differently according to the importance of surrounding pedestrians. Furthermore, Convolutional Neural Networks (CNN) is introduced into Social LSTM model to consider both the interaction between people and the characteristic of scene scale in the prediction process. Experiments are carried out against baseline methods, and the results demonstrated that combining Social LSTM with attention mechanism or CNN can improve the performance of pedestrian trajectory prediction. © 2022 IEEE.

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

Fuzzy Deep Forest With Deep Contours Feature for Leaf Cultivar Classification

引用

IEEE TRANSACTIONS ON FUZZY systems 2022年第12期30卷 5431-5444页

作者： Zheng, Wenbo Yan, Lan Gou, Chao Wang, Fei-Yue Wuhan Univ Technol Sch Comp Sci & Artificial Intelligence Wuhan 430070 Peoples R China Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Artificial Intelligence Beijing 100190 Peoples R China Sun Yat sen Univ Sch Intelligent Syst Engn Guangzhou 510275 Peoples R China Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Natl Univ Def Technol Res Ctr Mil Computat Expt & Parallel Syst Technol Changsha 410073 Peoples R China

Deep learning is a compelling technique for feature extraction due to its adaptive capacity of processing and providing deeper image information. However, for the task of leaf cultivar classification, the deep learning-based classifier model is unable to extract contour features of leaf images deeply due to the lack of large specialized datasets and expert knowledge annotations. Also, the scale/size of the current leaf cultivar dataset does not meet the needs of deep neural networks (DNNs). In particular, the high model complexity of DNNs implies that deep-learning-based neural networks seem to must require a large dataset to achieve good performance, but facing the fact that the leaf cultivar dataset often is small, even some classes in this kind of datasets contain less than ten images/examples. To overcome these problems and inspired by the resounding success of fuzzy logic, we propose a novel fuzzy ensemble model for leaf cultivar classification. To extract the contours of leaves, we first propose generative adversarial networks-based methods. Second, to improve the ability of feature representation, we present a data augmentation method to transform our contour features. Third, to get the essential features of leaves, we design a novel generation of the fuzzy random forest. Finally, to achieve accurate classification, we design a novel deep learning strategy, namely deep fuzzy representation learning, integrating and cascading a lot of our fuzzy random forests. Experimental results show that our model outperforms other existing state-of-the-arts on three real-world datasets, and performs much better than the original deep forest and DNN-based algorithms particularly.

关键词： Contour feature learning data augmentation deep forest fuzzy logic

来源：评论

学校读者我要写书评

暂无评论

TAO-Based Data Governance in intelligent Transportation systems

TAO-Based Data Governance in Intelligent Transportation Syst...

引用

International Conference on intelligent Transportation

作者： Xiaolong Liang Juanjuan Li Rui Qin Wenwen Ding Fei-Yue Wang Faculty of Innovation Engineering Macau University of Science and Technology Macao China State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing China The State Key Laboratory for Management and Control of Complex Systems and State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing China The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences and Collaborative Laboratory for Intelligent Science and Systems Beijing China

intelligent Transportation systems (ITS) are heavily dependent on private user data. However, most ITS fails to harness the potential of this invaluable data due to the absence of effective data governance mechanisms promoting users' contributions. Moreover, data contributors face security risks such as privacy preservation, data leakage, etc., as well as high costs in data sharing, while benefits are disproportionately reaped by system operators and interaction. This systemic imbalance could indirectly incentivize a surge in inactions and even malicious actions. In response to these challenges, we propose the design of a True Autonomous Organization (TAO) for ITS, namely ITS TAO. Utilizing the newly designed decision models with decentralized organization structures and the three-power structure, ITS TAO aims to realize the fair distribution of rights and benefits for ITS data contributors. Furthermore, we design a real-time evaluation system based on parallel intelligence capable of identifying potential hazards.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Recurrent Neural Network based Partially Observed Feedback control of Musculoskeletal Robots 7

Recurrent Neural Network based Partially Observed Feedback C...

引用

7th IEEE International Conference on Advanced Robotics and Mechatronics

作者： Chen, Jiahao Huang, Xiao Wang, Xiaona Qiao, Hong Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Beijing Inst Technol Sch Mechatron Engn Adv Innovat Ctr Intelligent Robots & Syst Beijing 100081 Peoples R China

ISBN: (数字)9781665483063

ISBN: (纸本)9781665483063

The musculoskeletal robot has become a promising research direction. However, the control problem of the musculoskeletal robot still limits its application. Especially in the real-world application, only some incomplete and imprecise feedback states could be obtained due to the limitation of sensors. Therefore, this paper proposes a recurrent neural network (RNN) based partially observed feedback control method of musculoskeletal robots. The RNN has the ability of working memory and can realize muscle control through implicitly inferring sufficient states from partially observed states. It can also generate muscle excitations with synergies to guarantee great generalization. The effectiveness of the proposed method is verified on a simulated musculoskeletal system. Compared with other deep reinforcement learning method, the proposed method achieves similar performance under sufficient feedback states and better performance under partially observed feedback states.

关键词： Deep learning Recurrent neural networks Mechatronics Neuromuscular Reinforcement learning Robot sensing systems Hardware

来源：评论

学校读者我要写书评

暂无评论

A feedback-based print quality improving strategy for FDM 3D printing: an optimal design approach

引用

INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY 2022年第3-4期120卷 2777-2791页

作者： Tamir, Tariku Sinshaw Xiong, Gang Fang, Qihang Dong, Xisong Shen, Zhen Wang, Fei-Yue Chinese Acad Sci Beijing Engn Res Ctr Intelligent Syst & Technol Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Artificial Intelligence Beijing 100049 Peoples R China Chinese Acad Sci Guangdong Engn Ctr 3D Printing & Intelligent Mfg Cloud Comp Ctr Dongguan 523808 Peoples R China Qingdao Acad Intelligent Ind Intelligent Mfg Ctr Qingdao 266109 Peoples R China

Fused deposition modelling (FDM) 3D printing, as a supporting technology in social manufacturing and cloud manufacturing, is a rapidly growing technology in the era of industry 4.0. It produces objects with a layer-by-layer material accumulation technique. However, qualitative uncertainties are the common challenges yet. In order to assure print quality, studying the error causing parameters and minimizing their effects are important. This paper presents a feedback-based error compensation strategy, which integrates a fuzzy inference system and a grey wolf optimization algorithm. The objectives are twofold. First, the possible errors in FDM 3D printing are discussed in detail and optimal error causing parameters are obtained in percentage. This is used to understand the effects of the printing errors in every phase of the 3D printing process. From the nine optimization configuration trials used, Config-6 that has 100 number of iterations and 60 wolves is selected due to its higher convergence speed and best fitness value. The integral absolute error (IAE) is used as an objective function and the global minimum is achieved in the iteration interval [86, 100]. The outputs of this optimization problem are used to achieve the next objective. Second, a closed-loop quality monitoring approach comprising of inner-loops and an outer-loop is taken. The three inner-loops are used to monitor the errors during pre-printing, printing, and post-printing, respectively. The outer-loop, on the other hand, is responsible for monitoring the aggregated errors in all the three 3D printing phases. The error compensation system simulation in Matlab is run for 10 s, and the results show that the "normal" range deformation factors are reached within less than 2 s for the inner-loops, whereas the outer-loop deformation factor is achieved within 2 s. The responses are within the acceptable time range.

关键词： Industry 4.0 3D printing Print quality Error causing parameters Deformations Fuzzy inference system

来源：评论

学校读者我要写书评

暂无评论

Transfer Methods for Vehicle Carbon Emission Models Based on the Parallel Transportation System

引用

IEEE TRANSACTIONS ON intelligent TRANSPORTATION systems 2024年第9期25卷 11664-11674页

作者： Sun, Yao Hu, Yunfeng Zhang, Hui Chen, Hong Wang, Fei-Yue Jilin Univ State Key Lab Automot Simulat & Control Changchun 130025 Peoples R China Jilin Univ Coll Commun Engn Changchun 130025 Peoples R China Beihang Univ Sch Transportat Sci & Engn Beijing 100091 Peoples R China Beihang Univ Ningbo Inst Technol NIT Ningbo 315323 Peoples R China Tongji Univ Coll Elect & Informat Engn Shanghai 200092 Peoples R China Macau Univ Sci & Technol Macau 999078 Peoples R China Chinese Acad Sci Beijing Engn Res Ctr Intelligent Syst & Technol Beijing 100098 Peoples R China Univ Chinese Acad Sci Sch Artificial Intelligence Beijing 100049 Peoples R China Chinese Acad Sci State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China

Vehicle carbon emission models are essential for monitoring and managing transportation emissions. Considering the numerous vehicle types and diverse driving conditions, modeling each vehicle individually is not possible. A generalized carbon emission module (GCEM) is proposed to realize fast model transfer across both light-duty gasoline and heavy-duty diesel vehicles. It has five main components: normalized torque function, category-specified fuel consumption rate function, denormalized function, emission-standard correction, and fuel-specified conversion. GCEM estimates carbon emissions based on the engine bench results of a reference vehicle, avoiding time-consuming and costly experiments. By normalizing engine torque, the significant power disparities within the same vehicle category can be effectively addressed. Integrating GCEM within the parallel transportation system is a feasible solution for monitoring complex traffic emissions. Through comprehensive comparisons, GCEM demonstrates a superior generalized ability in this data-limited scenario than three baseline models, both for various test vehicles and diverse driving conditions. As real-world traffic is a typical data-limited scenario, GCEM is a promising and practical transfer method for estimating vehicle carbon emissions.

关键词： Generalized carbon emission module modular integrated emission model parallel transportation system transfer method spatio-temporal characterization

来源：评论

学校读者我要写书评

暂无评论

Design and Optimization of a control Framework for Robot Assisted Additive Manufacturing Based on the Stewart Platform

引用

INTERNATIONAL JOURNAL OF control AUTOMATION AND systems 2022年第3期20卷 968-982页

作者： Tamir, Tariku Sinshaw Xiong, Gang Dong, Xisong Fang, Qihang Liu, Sheng Lodhi, Ehtisham Shen, Zhen Wang, Fei-Yue Chinese Acad Sci Beijing Engn Res Ctr Intelligent Syst & Technol Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Artificial Intelligence Beijing 100049 Peoples R China Chinese Acad Sci Cloud Comp Ctr Guangdong Engn Res Ctr 3D Printing & Intelligent Dongguan 523808 Peoples R China Qingdao Acad Intelligent Ind Intelligent Mfg Ctr Qingdao 266109 Peoples R China

Additive manufacturing, also known as 3D printing, is an emerging technology. The existing additive manufacturing technologies deploy a 3-axis printing mechanism where the material accumulation grows only in the z-direction. This results in limited printing freedom. Apart from this, support structures are needed to print overhang structures. Removal of these supports ultimately reduces print quality. This paper proposes a novel robot-assisted additive manufacturing along with a control system framework, which possesses multi-directional printing without support structures. Taking the advantage of its high stiffness and high payload-to-weight ratio, a 6-degree of freedom Stewart platform manipulator is designed to substitute the printer build plate. The kinematics and dynamics of the manipulator is formulated. Then, an extended proportion-derivation sliding mode controller is designed for trajectory tracking. The modified grey wolf optimization algorithm is applied to compute the optimal controller parameters. The integral absolute error (IAE) is used as a cost function and its minimum value is reached in the iteration interval [75,100]. The analytical model simulation in MATlab is run for 10 seconds, and the results show that the desired length trajectories of the six legs of the manipulator are achieved after 3.5 seconds. The performance of the analytical model is verified on the automated dynamic analysis of mechanical systems (ADAMS).

关键词： Additive manufacturing extended proportion-derivation sliding mode controller grey wolf optimization print quality Stewart platform

来源：评论

学校读者我要写书评

暂无评论

Deep Rank-Consistent Pyramid Model for Enhanced Crowd Counting

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING systems 2025年第1期36卷 299-312页

作者： Gao, Jiaqi Huang, Zhizhong Lei, Yiming Shan, Hongming Wang, James Z. Wang, Fei-Yue Zhang, Junping Fudan Univ Shanghai Key Lab Intelligent Informat Proc Sch Comp Sci Shanghai 200433 Peoples R China Fudan Univ Inst Sci & Technol Brain Inspired Intelligence Shanghai 200433 Peoples R China Fudan Univ MOE Frontiers Ctr Brain Sci Shanghai 200433 Peoples R China Shanghai Ctr Brain Sci & Brain Inspired Technol Shanghai 201210 Peoples R China Penn State Univ Coll Informat Sci & Technol University Pk PA 16802 USA Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Macau Univ Sci & Technol Inst Syst Engn Taipa Macao Peoples R China Macau Univ Sci & Technol Inst Syst Engn Int Acad Advisory Comm Taipa Macao Peoples R China Qingdao Acad Intelligent Ind Qingdao 266109 Peoples R China

Most conventional crowd counting methods utilize a fully-supervised learning framework to establish a mapping between scene images and crowd density maps. They usually rely on a large quantity of costly and time-intensive pixel-level annotations for training supervision. One way to mitigate the intensive labeling effort and improve counting accuracy is to leverage large amounts of unlabeled images. This is attributed to the inherent self-structural information and rank consistency within a single image, offering additional qualitative relation supervision during training. Contrary to earlier methods that utilized the rank relations at the original image level, we explore such rank-consistency relation within the latent feature spaces. This approach enables the incorporation of numerous pyramid partial orders, strengthening the model representation capability. A notable advantage is that it can also increase the utilization ratio of unlabeled samples. Specifically, we propose a Deep Rank-consist Ent pyrAmid Model (DREAM), which makes full use of rank consistency across coarse-to-fine pyramid features in latent spaces for enhanced crowd counting with massive unlabeled images. In addition, we have collected a new unlabeled crowd counting dataset, FUDAN-UCC, comprising 4000 images for training purposes. Extensive experiments on four benchmark datasets, namely UCF-QNRF, ShanghaiTech PartA and PartB, and UCF-CC-50, show the effectiveness of our method compared with previous semi-supervised methods. The codes are available at https://***/bridgeqiqi/DREAM.

关键词： Feature extraction Training Task analysis Head labeling Predictive models Magnetic heads Crowd counting feature pyramid ranking semi-supervised learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：