A5100390198- Advanced Graph Neural Networks
- Advanced Image and Video Retrieval Techniques
- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Graph Theory and Algorithms
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Topic Modeling
- Anomaly Detection Techniques and Applications
- Image Retrieval and Classification Techniques
- Image Processing Techniques and Applications
- Image and Signal Denoising Methods
- Bayesian Modeling and Causal Inference
- Advanced Image Processing Techniques
- Algorithms and Data Compression
- Advanced Measurement and Detection Methods
- Advanced Data Compression Techniques
- Natural Language Processing Techniques
- Neural Networks and Applications
- Multimodal Machine Learning Applications
- Machine Learning and Algorithms
- Digital Media Forensic Detection
- Robotics and Sensor-Based Localization
- Remote-Sensing Image Classification
- Text and Document Classification Technologies
The University of Adelaide
2015-2025
Australian Centre for Robotic Vision
2019-2025
Zhejiang Gongshang University
2023-2025
Shijiazhuang Tiedao University
2024
National University of Singapore
2009-2024
State Key Laboratory of Mobile Networks and Mobile Multimedia Technology
2024
Changsha University
2024
ZTE (United States)
2024
Shandong Xiehe University
2024
Huizhou University
2022-2023
In this paper, we revisit the joint probabilistic data association (JPDA) technique and propose a novel solution based on recent developments in finding m-best solutions to an integer linear program. The key advantage of approach is that it makes JPDA computationally tractable applications with high target and/or clutter density, such as spot tracking fluorescence microscopy sequences pedestrian surveillance footage. We also show our algorithm embedded simple framework surprisingly...
Human motion modeling is a classic problem in computer vision and graphics. Challenges human include high dimensional prediction as well extremely complicated dynamics.We present novel approach to based on convolutional neural networks (CNN). The hierarchical structure of CNN makes it capable capturing both spatial temporal correlations effectively. In our proposed approach, long-term encoder used encode the whole given sequence into hidden variable, which with decoder predict remainder...
The decoy-state scheme is the most widely implemented quantum-key-distribution protocol in practice. In order to account for finite-size key effects on achievable secret generation rate, a rigorous statistical fluctuation analysis required. Originally, heuristic Gaussian-approximation technique was used this purpose, which, despite its analytical convenience, not sufficiently rigorous. has recently been made by using Chernoff bound. There considerable gap, however, between key-rate bounds...
As an integral component of blind image deblurring, non-blind deconvolution removes blur with a given kernel, which is essential but difficult due to the ill-posed nature inverse problem. The predominant approach based on optimization subject regularization functions that are either manually designed or learned from examples. Existing learning-based methods have shown superior restoration quality not practical enough their restricted and static model design. They solely focus learning prior...
The feature matching problem is a fundamental in various areas of computer vision including image registration, tracking and motion analysis. Rich local representation key part efficient methods. However, when the features are limited to coordinate points, it becomes challenging extract rich representations. Traditional approaches use pairwise or higher order handcrafted geometric get robust matching; this requires solving NP-hard assignment problems. In paper, we address by proposing graph...
It is well known that path planning has always been an important study area for intelligent ships, especially unmanned surface vehicles (USVs). Therefore, it necessary to the path-planning algorithm USVs. As one of basic algorithms USV planning, rapidly-exploring random tree (RRT) popular due its simple structure, high speed and ease modification. However, also some obvious drawbacks problems. Designed perfect defects RRT improve performance USVs, enhanced proposed in this study, called...
This paper reviews the challenge on constrained high dynamic range (HDR) imaging that was part of New Trends in Image Restoration and Enhancement (NTIRE) workshop, held conjunction with CVPR 2022. manuscript focuses competition set-up, datasets, proposed methods their results. The aims at estimating an HDR image from multiple respective low (LDR) observations, which might suffer under-or over-exposed regions different sources noise. is composed two tracks emphasis fidelity complexity...
In recent years, we have witnessed a surge of Graph Neural Networks (GNNs), most which can learn powerful representations in an end-to-end fashion with great success many real-world applications. They resemblance to Probabilistic Graphical Models (PGMs), but break free from some limitations PGMs. By aiming provide expressive methods for representation learning instead computing marginals or likely configurations, GNNs flexibility the choice information flowing rules while maintaining good...
Time-varying problems are prevalent in engineering, presenting a significant challenge due to the fluctuations parameters and goals at different time points. The Zeroing Neural Network (ZNN), specialized form of Recurrent (RNN) developed by Zhang et al., has gained attention for its rapid convergence speed robustness making it valuable tool real-time solving diverse time-varying problems. This review article explores practical applications ZNN across various domains past two decades,...
Feature matching is a key problem in computer vision and pattern recognition. One way to encode the essential interdependence between potential feature matches cast as inference graphical model, though recently alternatives such spectral methods, or approaches based on convex-concave procedure have achieved state-of-the-art. Here we revisit use of models for matching, propose belief propagation scheme which exhibits following advantages: (1) explicitly enforce one-to-one constraints, (2)...
Exploiting label dependency for multi-label image classification can significantly improve performance. Probabilistic Graphical Models are one of the primary methods representing such dependencies. The structure graphical models, however, is either determined heuristically or learned from very limited information. Moreover, neither these approaches scales well to large complex graphs. We propose a principled way learn model by considering input features and labels, together with loss...
This paper is focused on the buckling and vibration analyses of microstructured structural elements, i.e., elements composed repetitive cells. The relationship between discrete equivalent nonlocal continuum specifically addressed from a numerical theoretical point view. beam considered herein modeled by some cells finite rigid segments elastic rotational springs. microstructure may come discreteness matter for small-scale structures (such as nanotechnology applications), but it can also be...
Abstract Perioperative administration of tranexamic acid (TXA) is thought to be related decreased postoperative implant-associated infection rates; however, the relationship remains unclear. We explored inhibitory effect TXA on both in vitro and vivo. investigated biofilm formation after through different detection methods, all which showed that reduces was further proven associated with protein polysaccharide contents biofilms. observed implants bacteria area strengthened neutrophil...
A continuous model is derived for the dynamics of two immiscible fluids with moving contact lines and insoluble surfactants based on thermodynamic principles. The continuum consists Navier-Stokes equations a convection-diffusion equation evolution surfactant fluid interface. interface condition, boundary condition slip velocity, dynamic angle are from consideration energy dissipations. Different types dissipations, including viscous dissipation, dissipations solid wall at line, as well...
Patch-based models that combine local image features or regions into loose geometric assemblies are a powerful paradigm for visual object tracking, and they present favorable properties such as robustness to partial occlusion, deformation, the ability address viewpoint changes. However, effectively exploiting spatial-temporal confidence scores of each patch construct robust tracker while ensuring low computational cost with dense discrete search remains challenging problem. In this paper, we...
Combination therapy, which involves the use of multiple drugs, has emerged as a promising approach to cancer treatment. However, traditional combination therapy development is constrained by vast experimental design space, requiring exhaustive testing drug ratios, concentrations, and encapsulation strategies. In this study, we present computational intelligence method combining active learning fine-grid optimization predict efficacy combinations, focusing on dual-drug-loaded polymeric...
This paper focuses on the problem that relevance feedback schemes based support vector machines (RF-SVM) always give a poor performance when numbers of positive/negative examples are strongly asymmetric. To address this issue, we propose random sampling SVM query expansion for learning. Firstly, adopt method to construct multiple asymmetric bagging classifiers (hard or binary each) and aggregate them form compound classifier by committee voting. Subsequently, voting results combined with...