A5100439115- Service-Oriented Architecture and Web Services
- Advanced Neural Network Applications
- Mobile Ad Hoc Networks
- Advanced Computational Techniques and Applications
- Generative Adversarial Networks and Image Synthesis
- Mobile Crowdsensing and Crowdsourcing
- Privacy-Preserving Technologies in Data
- Educational Reforms and Innovations
- Advanced Steganography and Watermarking Techniques
- Domain Adaptation and Few-Shot Learning
- Blockchain Technology Applications and Security
- Cloud Computing and Resource Management
- Data Management and Algorithms
- Semantic Web and Ontologies
- Advanced Image and Video Retrieval Techniques
- Human Pose and Action Recognition
- Image Retrieval and Classification Techniques
- Anomaly Detection Techniques and Applications
- Opportunistic and Delay-Tolerant Networks
- Medical Image Segmentation Techniques
- Energy Efficient Wireless Sensor Networks
- Caching and Content Delivery
- Digital Media Forensic Detection
- Advanced Image Processing Techniques
- Video Surveillance and Tracking Methods
Inha University
2016-2025
Chongqing University of Technology
2009-2024
Hebei University
2012-2023
Shanghai Institute of Technology
2023
Yunnan Normal University
2023
China Academy of Information and Communications Technology
2023
Lanzhou University of Finance and Economics
2023
Nanchang Institute of Technology
2013-2022
PLA Academy of Military Science
2022
Intel (United States)
2022
Automatic assessing the location and extent of liver tumor is critical for radiologists, diagnosis clinical process. In recent years, a large number variants U-Net based on Multi-scale feature fusion are proposed to improve segmentation performance medical image segmentation. Unlike previous works which extract context information via applying multi-scale fusion, we propose novel network named Attention Net (MA-Net) by introducing self-attention mechanism into our method adaptively integrate...
The upgraded version of intelligent image-activated cell sorting (iIACS) has enabled higher-throughput and more sensitive image-based single live cells from heterogeneous populations.
Most metric learning techniques typically focus on sample embedding learning, while implicitly assume a homogeneous local neighborhood around each sample, based the metrics used in training ( e.g., hypersphere for Euclidean distance or unit hyperspherical crown cosine distance). As real-world data often lies low-dimensional manifold curved high-dimensional space, it is unlikely that everywhere of shares same structures input space. Besides, considering non-linearity neural networks,...
Web services technology has received much attention in the last few years, and a lot of research efforts have been devoted to utilizing on Internet fulfill consumers' requirements. However, little done current status web Internet, which great impact research. Enlightened by this situation, we made an exploratory study Internet. Our mainly focused investigation four aspects, including number, complexity, quality description function diversity available A system is built up harvest from...
Abstract We consider medical image transformation problems where a grayscale is transformed into color image. The colorized should have the same features as input because extra synthesized can increase possibility of diagnostic errors. In this paper, to secure images and improve quality images, well leverage unpaired training data, colorization network proposed based on cycle generative adversarial (CycleGAN) model, combining perceptual loss function total variation (TV) function. Visual...
Purpose To improve hyperpolarized 13 C (HP‐ C) MRI by image denoising with a new approach, patch‐based higher‐order singular value decomposition (HOSVD). Methods The benefit of using HOSVD method to denoise dynamic HP‐ MR imaging data was investigated. Image quality and the accuracy quantitative analyses following were evaluated first simulated [1‐ C]pyruvate its metabolic product, C]lactate, compared results global method. then applied healthy volunteer HP EPI studies. Voxel‐wise kinetic...
Accurately segmenting polyps from colonoscopy images is essential for diagnosing colorectal cancer. Despite the tremendous success of deep convolutional neural networks in automatic polyp segmentation, it suffers domain shift issues, where trained model yields performance deterioration on unseen test datasets. This paper proposes an illumination enhancement-based generalization approach to improve capability datasets and alleviate this issue. In particular, image decomposition module (IDM)...
Most of existing signal modulation recognition methods attempt to establish a machine learning mechanism by training with large number annotated samples, which is hardly applied the real-world electronic reconnaissance scenario where only few samples can be intercepted in advance. Few-Shot Learning (FSL) aims learn from classes lot and transform knowledge support thus realizing model generalization. In this paper, novel FSL framework called Attention Relation Network (ARN) proposed,...
In clinical practice, medical image analysis has played a key role in disease diagnosis. One of the important steps is to perform an accurate organ or tissue segmentation for assisting professionals making correct diagnoses. Despite tremendous progress deep learning-based approaches, they often fail generalize test datasets due distribution discrepancies across domains. Recent advances aligning domain gaps by using bi-directional GANs (e.g., CycleGAN) have shown promising results, but strict...
Multiplicative noise, also known as speckle is signal dependent and difficult to remove. Based on a fourth‐order PDE model, this paper proposes novel approach remove the multiplicative noise images. In practice, Fourier transform logarithm strategy are utilized noisy image convert convolutional into additive so that can be removed by using traditional removal algorithm in frequency domain. For removal, new model developed, which avoids blocky effects produced second‐order attains better...
To protect the copyright of digital image, this paper proposed a combined Discrete Wavelet Transform (DWT) and Cosine (DCT) based watermarking scheme. embed watermark, cover image was decomposed by 2-level DWT, HL2 sub-band coefficient divided into 4x4 blocks, then DCT performed on each these blocks. The watermark bit embedded predefined pattern_0 or pattern_1 middle band coefficients DCT. After insertion, inverse applied to blocks coefficient, DWT obtain watermarked image. For extraction,...
E-wallets have started to grow in popularity, reaching a tipping point some countries. This can be attributed the worldwide use of payment-enabled devices and ubiquity e-wallet acceptance by larger smaller retailers. As more customers adopt e-wallets they may also become big target cybercrime. facilitates financial transactions via smartphones which is lucrative opportunity for cybercriminals. paper presents security assessment Android apps provided Canada's leading banks.
Three-dimensional images are frequently used in medical imaging research for classification, segmentation, and detection. However, the limited availability of 3D hinders progress due to network training difficulties. Generative methods have been proposed create using AI techniques. Nevertheless, 2D approaches difficulty dealing with anatomical structures, which can result discontinuities between slices. To mitigate these discontinuities, several generative networks proposed. scarcity...
As an extremely powerful probability model, Gaussian mixture model (GMM) has been widely used in fields of pattern recognition, information processing and data mining. If the number Gaussians is pre-known, well-known Expectation-Maximization (EM) algorithm could be to estimate parameters model. However, many practical applications, components not known.Then modeling becomes a compound problem determination parameter estimation for mixture, which rather difficult. In this paper, we propose...
Recent smartphones are equipped with various sensors, such as an accelerometer, GPS, and a gravity sensor, have high-performance wireless communication capabilities. Through the ubiquitous presence of powerful mobile devices, crowdsensing lets ordinary people collectively gather share real-time multimedia data. Multimedia has made large-scale participatory sensing viable in speedy cost-efficient manner, but it also introduces some security privacy concerns. Personally identifiable...
N/A Type of Report: MS Project Report Department Computer Science & Engineering Washington University in St. Louis Campus Box 1045 Louis, MO 63130 ph: (314) 935-6160 A Survey on Communication Networks Emergency Warning Systems Student: Yan Li Advisor: Raj Jain
Recently, outfit compatibility modeling, which aims to evaluate the of a given that comprises set fashion items, has gained growing research attention. Although existing studies have achieved prominent progress, most them overlook essential global representation learning, and hidden complementary factors behind uncovering. Towards this end, we propose an Outfit Compatibility Modeling scheme via Complementary Factorization, termed as OCM-CF. In particular, OCM-CF consists two key components:...