A5079217067- Meteorological Phenomena and Simulations
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Energy Load and Power Forecasting
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Climate variability and models
- Advanced Vision and Imaging
- Hydrological Forecasting Using AI
- Video Analysis and Summarization
- Solar Radiation and Photovoltaics
- Multimodal Machine Learning Applications
- Advanced Neural Network Applications
- Gait Recognition and Analysis
- Air Quality Monitoring and Forecasting
- Remote-Sensing Image Classification
- Infrared Target Detection Methodologies
- 3D Shape Modeling and Analysis
- Parallel Computing and Optimization Techniques
- Music and Audio Processing
- Face recognition and analysis
- Flood Risk Assessment and Management
- Urban Heat Island Mitigation
- Image and Signal Denoising Methods
Third Affiliated Hospital of Zhengzhou University
2025
Chinese Academy of Sciences
2009-2024
James Madison University
2023-2024
Scripps Institution of Oceanography
2022-2024
University of California, San Diego
2022-2024
UC San Diego Health System
2024
Pennsylvania State University
2018-2024
Institute of Automation
2005-2023
ShanghaiTech University
2022-2023
University of Chinese Academy of Sciences
2020-2023
Human identification at a distance has recently gained growing interest from computer vision researchers. Gait recognition aims essentially to address this problem by identifying people based on the way they walk. In paper, simple but efficient gait algorithm using spatial-temporal silhouette analysis is proposed. For each image sequence, background subtraction and correspondence procedure are first used segment track moving silhouettes of walking figure. Then, eigenspace transformation...
Graph convolutional networks (GCNs) have been widely used and achieved remarkable results in skeleton-based action recognition. In GCNs, graph topology dominates feature aggregation therefore is the key to extracting representative features. this work, we propose a novel Channel-wise Topology Refinement Convolution (CTR-GC) dynamically learn different topologies effectively aggregate joint features channels for The proposed CTR-GC models channel-wise through learning shared as generic prior...
Analysis of motion patterns is an effective approach for anomaly detection and behavior prediction. Current approaches the analysis depend on known scenes, where objects move in predefined ways. It highly desirable to automatically construct object which reflect knowledge scene. In this paper, we present a system learning prediction based proposed algorithm robustly tracking multiple objects. algorithm, foreground pixels are clustered using fast accurate fuzzy K-means algorithm. Growing...
The Visual Object Tracking challenge VOT2017 is the fifth annual tracker benchmarking activity organized by VOT initiative. Results of 51 trackers are presented; many state-of-the-art published at major computer vision conferences or journals in recent years. evaluation included standard and other popular methodologies a new "real-time" experiment simulating situation where processes images as if provided continuously running sensor. Performance tested typically far exceeds baselines. source...
Network intrusion detection aims at distinguishing the attacks on Internet from normal use of Internet. It is an indispensable part information security system. Due to variety network behaviors and rapid development attack fashions, it necessary develop fast machine-learning-based algorithms with high rates low false-alarm rates. In this correspondence, we propose algorithm based AdaBoost algorithm. algorithm, decision stumps are used as weak classifiers. The rules provided for both...
In this paper, we proposed a hierarchical clustering framework to classify vehicle motion trajectories in real traffic video based on their pairwise similarities. First raw are pre-processed and resampled at equal space intervals. Then spectral is used group with similar spatial patterns. Dominant paths lanes can be distinguished as result of two-layer clustering. Detection novel also possible the results. Experimental results demonstrate superior performance compared conventional fuzzy...
Visual tracking usually involves an optimization process for estimating the motion of object from measured images in a video sequence. In this paper, new evolutionary approach, PSO (particle swarm optimization), is adopted visual tracking. Since dynamic problem which simultaneously influenced by state and time, we propose sequential particle framework incorporating temporal continuity information into traditional algorithm. addition, parameters are changed adaptively according to fitness...
Object appearance modeling is crucial for tracking objects, especially in videos captured by nonstationary cameras and reasoning about occlusions between multiple moving objects. Based on the log-euclidean Riemannian metric symmetric positive definite matrices, we propose an incremental subspace learning algorithm which covariance matrices of image features are mapped into a vector space with metric. algorithm, develop block-division model captures both global local spatial layout...
The one-shot multi-object tracking, which integrates object detection and ID embedding extraction into a unified network, has achieved groundbreaking results in recent years. However, current trackers solely rely on single-frame detections to predict candidate bounding boxes, may be unreliable when facing disastrous visual degradation, e.g., motion blur, occlusions. Once target box is mistakenly classified as background by the detector, temporal consistency of its corresponding tracklet will...
Many scientific problems require multiple distinct computational tasks to be executed in order achieve a desired solution. We introduce the Ensemble Toolkit (EnTK) address challenges of scale, diversity and reliability they pose. describe design implementation EnTK, characterize its performance integrate it with two exemplar use cases: seismic inversion adaptive analog ensembles. perform nine experiments, characterizing EnTK overheads, strong weak scalability, case imple-mentations, at scale...
High-level semantic understanding of vehicle motion behaviors is often based on trajectory clustering. In this paper, we propose an effective clustering framework in which a coarse-to-fine strategy taken. Our consists four stages: smoothing, feature extraction, coarse and fine Wavelet decomposition imposed raw trajectories to reduce noise the smoothing stage. Besides commonly used positional feature, novel called directional histogram proposed describe statistic distribution extraction Both...
In this paper, we present a framework for active contour-based visual tracking using level sets. The main components of our include initialization, color-based contour evolution, adaptive shape-based evolution non-periodic motions, dynamic periodic and the handling abrupt motions. For initialization tracking, develop an optical flow-based algorithm automatically initializing contours at first frame. Markov random field theory is used to measure correlations between values neighboring pixels...
The rising temperature is one of the key indicators a warming climate, capable causing extensive stress to biological systems as well built structures.Ambient collected at ground level can have higher variability than regional weather forecasts, which fail capture local dynamics. There remains clear need for accurate air prediction suburban scale high temporal and spatial resolutions. This research proposed framework based on long short-term memory (LSTM) deep learning network generate...
Abstract Reliably quantifying uncertainty in precipitation forecasts remains a critical challenge. This work examines the application of deep learning (DL) architecture, Unet, for postprocessing deterministic numerical weather predictions to improve their skills and deriving forecast uncertainty. Daily accumulated 0–4-day are generated from 34-yr reforecast based on West Weather Research Forecasting (West-WRF) mesoscale model, developed by Center Western Water Extremes. The Unet learns...
Cigarette smoking has been linked to severe and persistent sleep disturbances alongside notable fluctuations in neuropeptide levels. Substance P (SubP), influenced by smoking, also impacts sleep-wake cycles. However, its specific role smoking-induced disorders remains unclear. This study aimed explore the connection between cigarette quality examining SubP levels cerebrospinal fluid (CSF) identifying potential treatment avenues for disorders. A total of 146 Chinese men (93 nonsmokers, 53...
In the field of MLLM-based GUI agents, compared to smartphones, PC scenario not only features a more complex interactive environment, but also involves intricate intra- and inter-app workflows. To address these issues, we propose hierarchical agent framework named PC-Agent. Specifically, from perception perspective, devise an Active Perception Module (APM) overcome inadequate abilities current MLLMs in perceiving screenshot content. From decision-making handle user instructions...
The authors propose a simple but efficient approach to gait recognition. For each image sequence, an improved background subtraction procedure is first used accurately extract spatial silhouettes of walker from the background. Then, eigenspace transformation time-varying silhouette shapes performed realize feature extraction. nearest neighbor classifier using spatio-temporal correlation or normalized Euclidean distance measure finally utilized in lower-dimensional for recognition, and some...
Activity analysis and semantic interpretation of tracked targets in a dynamic image sequence has recently attracted more attentions computer vision. In this paper, framework for vehicle pedestrian behaviors is proposed practical applications visual traffic surveillance. The trajectories recorded the tracking process are analyzed using clustering classification on which high level based Experimental results presented to illustrate performance algorithm.
Most existing active learning approaches are supervised. Supervised has the following problems: inefficiency in dealing with semantic gap between distribution of samples feature space and their labels, lack ability selecting new that belong to categories have not yet appeared training samples, adaptability changes interpretation sample categories. To tackle these problems, we propose an unsupervised framework based on hierarchical graph-theoretic clustering. In framework, two promising...
Large variations in image background may cause partial matching and normalization problems for histogram-based representations, i.e., the histograms of same category have bins which are significantly different, produce large changes differences between corresponding bins. In this paper, we deal with problem by using ratios bin values histograms, rather than values' used traditional histogram distances. We propose a ratio-based distance (BRD), is an intra-cross-bin distance, contrast previous...