A5047405956- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Advanced Vision and Imaging
- Autonomous Vehicle Technology and Safety
- Human Pose and Action Recognition
- Robotics and Sensor-Based Localization
- Multimodal Machine Learning Applications
- Domain Adaptation and Few-Shot Learning
- Anomaly Detection Techniques and Applications
- Visual Attention and Saliency Detection
- Image Retrieval and Classification Techniques
- Robotic Path Planning Algorithms
- Speech and Audio Processing
- Neural Networks and Applications
- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Advanced Data Compression Techniques
- Direction-of-Arrival Estimation Techniques
- Face recognition and analysis
- Blind Source Separation Techniques
- Image Processing Techniques and Applications
- Face and Expression Recognition
- Image Enhancement Techniques
- Hand Gesture Recognition Systems
Xi'an Jiaotong University
2016-2025
Institute of Art
2015-2024
National Engineering Research Center for Information Technology in Agriculture
2024
Microsoft (United States)
2023
Shandong Institute of Automation
2014-2023
Chinese Academy of Sciences
2014-2023
Beijing University of Posts and Telecommunications
2023
Communication University of China
2023
Beijing Academy of Artificial Intelligence
2002-2020
Tsinghua University
2020
In this paper, we study the salient object detection problem for images. We formulate as a binary labeling task where separate from background. propose set of novel features, including multiscale contrast, center-surround histogram, and color spatial distribution, to describe locally, regionally, globally. A conditional random field is learned effectively combine these features detection. Further, extend proposed approach detect sequential images by introducing dynamic features. collected...
"Computer vision and pattern recognition." , 84(9), pp. 1265–1266 Additional informationNotes on contributorsNanning Zheng Email: nnzheng@mail.xjtu.edu.cn George Loizou george@dcs.bbk.ac.uk Xuguang Lan xglan@aiar.xjtu.edu.cn Xuelong Li xuelong_li@ieee.org
In this paper, we formulate the stereo matching problem as a Markov network and solve it using Bayesian belief propagation. The consists of three coupled random fields that model following: smooth field for depth/disparity, line process depth discontinuity, binary occlusion. After eliminating by introducing two robust functions, apply propagation algorithm to obtain maximum posteriori (MAP) estimation in network. Other low-level visual cues (e.g., image segmentation) can also be easily...
Person re-identification across cameras remains a very challenging problem, especially when there are no overlapping fields of view between cameras. In this paper, we present novel multi-channel parts-based convolutional neural network (CNN) model under the triplet framework for person re-identification. Specifically, proposed CNN consists multiple channels to jointly learn both global full-body and local body-parts features input persons. The is trained by an improved loss function that...
Salient object detection has been attracting a lot of interest, and recently various heuristic computational models have designed. In this paper, we regard saliency map computation as regression problem. Our method, which is based on multi-level image segmentation, uses the supervised learning approach to regional feature vector score, finally fuses scores across multiple levels, yielding map. The contributions lie in two-fold. One that show our approach, integrates contrast, property...
We study visual attention by detecting a salient object in an input image. formulate detection as image segmentation problem, where we separate the from background. propose set of novel features including multi-scale contrast, center-surround histogram, and color spatial distribution to describe locally, regionally, globally. A conditional random field is learned effectively combine these for detection. also constructed large database containing tens thousands carefully labeled images...
As a robust nonlinear similarity measure in kernel space, correntropy has received increasing attention domains of machine learning and signal processing. In particular, the maximum criterion (MCC) recently been successfully applied regression filtering. The default function is Gaussian kernel, which is, course, not always best choice. this work, we propose generalized that adopts density (GGD) as (not necessarily Mercer kernel), present some important properties. We further (GMCC), apply it...
Skeleton-based human action recognition has recently attracted increasing attention due to the popularity of 3D skeleton data. One main challenge lies in large view variations captured actions. We propose a novel adaptation scheme automatically regulate observation viewpoints during occurrence an action. Rather than re-positioning skeletons based on defined prior criterion, we design adaptive recurrent neural network (RNN) with LSTM architecture, which enables itself adapt most suitable from...
Skeleton-based human action recognition has attracted great interest thanks to the easy accessibility of skeleton data. Recently, there is a trend using very deep feedforward neural networks model 3D coordinates joints without considering computational efficiency. In this paper, we propose simple yet effective semantics-guided network (SGN) for skeleton-based recognition. We explicitly introduce high level semantics (joint type and frame index) into enhance feature representation capability....
In crowd scenarios, reliable trajectory prediction of pedestrians requires insightful understanding their social behaviors. These behaviors have been well investigated by plenty studies, while it is hard to be fully expressed hand-craft rules. Recent studies based on LSTM networks shown great ability learn However, many these methods rely previous neighboring hidden states but ignore the important current intention neighbors. order address this issue, we propose a data-driven state...
Skeleton-based human action recognition has recently attracted increasing attention thanks to the accessibility and popularity of 3D skeleton data. One key challenges in lies large variations representations when they are captured from different viewpoints. In order alleviate effects view variations, this paper introduces a novel adaptation scheme, which automatically determines virtual observation viewpoints over course an learning based data driven manner. Instead re-positioning skeletons...
We propose a novel automatic salient object segmentation algorithm which integrates both bottom-up stimuli and object-level shape prior, i.e., has well-defined closed boundary. Our approach is formalized as an iterative energy minimization framework, leading to binary of the object. Such initialized with saliency map computed through context analysis based on multi-scale superpixels. Object-level prior then extracted combining boundary information. Both update after each iteration....
The steady-state excess mean square error (EMSE) of the adaptive filtering under maximum correntropy criterion (MCC) has been studied. For Gaussian noise case, we establish a fixed-point equation to solve exact value EMSE, while for non-Gaussian derive an approximate analytical expression based on Taylor expansion approach. Simulation results agree with theoretical calculations quite well.
We propose a Bayesian approach to image hallucination. Given generic low resolution image, we hallucinate high using set of training images. Our work is inspired by recent progress on natural statistics that the priors primitives can be well represented examples. Specifically, primal sketch (e.g., edges, ridges and corners) are constructed used enhance quality hallucinated image. Moreover, contour smoothness constraint enforces consistency in Markov-chain based inference algorithm. A...
Pose variation remains one of the major factors that adversely affect accuracy person re-identification. Such is not arbitrary as body parts (e.g. head, torso, legs) have relative stable spatial distribution. Breaking down variability global appearance regarding distribution potentially benefits matching. We therefore learn a novel similarity function, which consists multiple sub-similarity measurements with each taking in charge subregion. In particular, we take advantage recently proposed...
The coronavirus disease 2019 (COVID-19) breaking out in late December is gradually being controlled China, but it still spreading rapidly many other countries and regions worldwide. It urgent to conduct prediction research on the development spread of epidemic. In this article, a hybrid artificial-intelligence (AI) model proposed for COVID-19 prediction. First, as traditional epidemic models treat all individuals with having same infection rate, an improved susceptible–infected (ISI)...
The maximum correntropy criterion (MCC) has received increasing attention in signal processing and machine learning due to its robustness against outliers (or impulsive noises). Some gradient based adaptive filtering algorithms under MCC have been developed available for practical use. fixed-point are, however, seldom studied. In particular, too little paid the convergence issue of algorithms. this letter, we will study problem give a sufficient condition guarantee algorithm.
Interest in autonomous driving (AD) and intelligent vehicles (IVs) is growing at a rapid pace due to the convenience, safety, economic benefits. Although number of surveys have reviewed research achievements this field, they are still limited specific tasks, lack systematic summary directions future. Here we propose Survey Surveys (SoS) for total technologies AD IVs that reviews history, summarizes milestones, provides perspectives, ethics, future directions. To our knowledge, article first...
In this paper, we study how to test the intelligence of an autonomous vehicle. Comprehensive testing is crucial both vehicle manufactories and customers. Existing approaches can be categorized into two kinds: scenario-based functionality-based testing. We first discuss shortcomings these kinds approaches, then propose a new framework combine benefits them. Based on semantic diagram definition for vehicles, explain design task evaluate results. Experiments show that approach provides quantitative way
The emerging development of connected and automated vehicles imposes a significant challenge on current vehicle control transportation systems. This paper proposes novel unified approach, Parallel Driving, cloud-based cyberphysical-social systems U+0028 CPSS U+0029 framework aiming at synergizing driving. study first introduces the ACP-based intelligent machine Then parallel driving is proposed in cyber-physical-social space, considering interactions among vehicles, human drivers,...
Conventional error correction codes (ECCs), such as the commonly used BCH code, have become increasingly inadequate for solid state drives (SSDs) capacity of NAND flash memory continues to increase and its reliability degrade. It is highly desirable deploy a much more powerful ECC, low-density parity-check (LDPC) significantly improve SSDs. Although LDPC code has had success in commercial hard disk drives, fully exploit capability SSDs demands unconventional fine-grained sensing, leading an...
In this paper, we address the person re-identification problem, discovering correct matches for a probe image from set of gallery images. We follow learning-to-rank methodology and learn similarity function to maximize difference between scores matched unmatched images same person. introduce at least three contributions re-identification. First, present an explicit polynomial kernel feature map, which is capable characterizing information all pairs patches two images, called...