A5082787229- Face and Expression Recognition
- Sparse and Compressive Sensing Techniques
- Face recognition and analysis
- Advanced Image and Video Retrieval Techniques
- Remote-Sensing Image Classification
- Image and Video Stabilization
- Image Processing Techniques and Applications
- Image and Signal Denoising Methods
- Image Retrieval and Classification Techniques
- Traditional and Medicinal Uses of Annonaceae
- Synthesis and biological activity
- Machine Learning and ELM
- Natural product bioactivities and synthesis
- Video Surveillance and Tracking Methods
- Domain Adaptation and Few-Shot Learning
- Advanced Image Processing Techniques
- Synthesis and Biological Evaluation
- Biometric Identification and Security
- Enzyme function and inhibition
- Carbon dioxide utilization in catalysis
- Phytochemistry and Biological Activities
- Fungal Plant Pathogen Control
- Smart Grid and Power Systems
- Indoor and Outdoor Localization Technologies
- Advanced Graph Neural Networks
University of South China
2023-2024
Guangdong Polytechnic Normal University
2014-2023
China Electric Power Research Institute
2021
North China Electric Power University
2021
Harbin Institute of Technology
2012-2018
Minjiang University
2017-2018
Cloud Computing Center
2012-2016
Nankai University
2009-2015
Collaborative Innovation Center of Chemical Science and Engineering Tianjin
2015
Lanzhou University
2009-2015
Locality and label information of training samples play an important role in image classification. However, previous dictionary learning algorithms do not take the locality atoms into account together process, thus their performance is limited. In this paper, a discriminative algorithm, called locality-constrained embedding (LCLE-DL) was proposed for First, preserved using graph Laplacian matrix learned instead conventional one derived from samples. Then, term constructed classification...
During the past several years, as one of most successful applications sparse coding and dictionary learning, dictionary-based face recognition has received significant attention. Although some surveys learning have been reported, there is no specialized survey concerning algorithms for recognition. This paper provides a To provide comprehensive overview, we not only categorize existing but also present details each category. Since number atoms an important impact on classification...
Both interclass variances and intraclass similarities are crucial for improving the classification performance of discriminative dictionary learning (DDL) algorithms. However, existing DDL methods often ignore combination between properties atoms coding coefficients. To address this problem, in paper, we propose a Fisher embedding (DFEDL) algorithm that simultaneously establishes models on learned Specifically, first construct atom model by exploring criterion atoms, which encourages same...
Recently, image prior learning has emerged as an effective tool for denoising, which exploits knowledge to obtain sparse coding models and utilize them reconstruct the clean from noisy one. Albeit promising, these prior-learning based methods suffer some limitations such lack of adaptivity failed attempts improve performance efficiency simultaneously. With purpose addressing problems, in this paper, we propose a Pyramid Guided Filter Network (PGF-Net) integrated with pyramid-based neural...
Supervised deep learning techniques have been widely explored in real photograph denoising and achieved noticeable performances. However, being subject to specific training data, most current image algorithms can easily be restricted certain noisy types exhibit poor generalizability across testing sets. To address this issue, we propose a novel flexible well-generalized approach, coined as dual meta attention network (DMANet). The DMANet is mainly composed of cascade the self-meta blocks...
Real photograph denoising is extremely challenging in low-level computer vision since the noise sophisticated and cannot be fully modeled by explicit distributions. Although deep-learning techniques have been actively explored for this issue achieved convincing results, most of networks may cause vanishing or exploding gradients, usually entail more time memory to obtain a remarkable performance. This article overcomes these challenges presents novel network, namely, PID controller guide...
With the development of deep learning technologies, recent research on real-world noisy image denoising has achieved a considerable improvement in performance. However, common limitation for existing approaches is imbalanced trade-off between accuracy and efficiency. To address this problem, we propose robust efficient denoiser, called hierarchical-based PID-attention network (HPDNet), to flexibly deal with sophisticated noise. The core our algorithm PID-attentive recurrent (PAR-Net) whose...
The lunar crater recognition plays a key role in exploration. Traditional methods are mainly based on the human observation that is usually combined with classical machine learning methods. These have some drawbacks, such as lacking objective criterion. Moreover, they can hardly achieve desirable results small or overlapping craters. To address these problems, we propose new convolutional neural network termed effective residual U-Net (ERU-Net) to recognize craters from digital elevation...
Recently, deep learning techniques are soaring and have shown dramatic improvements in real-world noisy image denoising. However, the statistics of real noise generally vary with different camera sensors in-camera signal processing pipelines. This will induce problems most denoisers for overfitting or degrading performance due to discrepancy between training test sets. To remedy this issue, we propose a novel flexible adaptive denoising network, coined as FADNet. Our FADNet is equipped plane...
Graph convolutional network (GCN) is an efficient for learning graph representations. However, it costs expensive to learn the high-order interaction relationships of node neighbor. In this paper, we propose a novel model and fuse multihop neighbor information relationships. We adopt weight-sharing mechanism design different order convolutions avoiding potential concerns overfitting. Moreover, new fusion (MIF) operator which mixes features from 1-hop k-hops. theoretically analyse...
In transfer learning model, the source domain samples and target usually share same class labels but have different distributions. general, existing algorithms ignore interclass differences intraclass similarities across domains. To address these problems, this article proposes a algorithm based on discriminative Fisher embedding adaptive maximum mean discrepancy (AMMD) constraints, called dictionary (DFEDTL). First, combining label information of part domain, we construct model to preserve...
Heterogeneous copper-in-charcoal-catalyzed click synthesis in 96-well polypropylene filter plates is an efficient method for the rapid generation of sufficient pure 2-alkoxyl-2-(1,2,3-triazole-1-yl) acetamide derivatives library by simple filtration, which directly assay products larvicidal activity against mosquitoes. In this procedure, copper nanoparticles on charcoal were arrayed into each well a plate, reagents delivered using pipette gun, and constant temperature shaker bath was used to...
Two new steroids, vladimuliecins A (1) and B (2), were isolated by bioassay-guided fractionation from the rhizome of Vladimiria muliensis. Compounds 1 2 are first examples possessing a pentacyclic 3α,5α-cyclopregnane-type framework. The structures (2) their deacetylated derivative (3) determined on basis IR, MS, 1D NMR, 2D X-ray data analyses. absolute configuration 10 stereogenic centers compounds was to be 3R,5R,6R,8S,9R,10R,13R,14R,17S,20R means auxiliary chiral MTPA derivatives optical...
The sparse representation classification (SRC) method proposed by Wright et al. is considered as the breakthrough of face recognition because its good performance. Nevertheless it still cannot perfectly address problem. main reason for this that variation poses, facial expressions, and illuminations image can be rather severe number available images are fewer than dimensions image, so a certain linear combination all training samples not able to fully represent test sample. In study, we...