A5052847739- 3D Shape Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Advanced Numerical Analysis Techniques
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Image and Object Detection Techniques
- Advanced Vision and Imaging
- COVID-19 diagnosis using AI
- Multimodal Machine Learning Applications
- Biometric Identification and Security
- AI in cancer detection
- Advanced Image and Video Retrieval Techniques
- Augmented Reality Applications
- Video Surveillance and Tracking Methods
- Radiomics and Machine Learning in Medical Imaging
- Robotics and Sensor-Based Localization
- Additive Manufacturing and 3D Printing Technologies
- Advanced Authentication Protocols Security
- Wood Treatment and Properties
- Advanced X-ray and CT Imaging
- Image and Signal Denoising Methods
- Face and Expression Recognition
- User Authentication and Security Systems
- 3D Surveying and Cultural Heritage
- Image Enhancement Techniques
University of Hull
2015-2024
University of Birmingham
2023
Yue Bei People's Hospital
2017-2022
Shantou University
2017-2022
Shanghai Institute for Science of Science
2022
China Agricultural University
2021
Umbo Computer Vision (United Kingdom)
2019
Northeast Forestry University
2016
Deep learning has been widely used in medical image segmentation and other aspects. However, the performance of existing models limited by challenge obtaining sufficient high-quality labeled data due to prohibitive annotation cost. To alleviate this limitation, we propose a new text-augmented model LViT (Language meets Vision Transformer). In our model, text is incorporated compensate for quality deficiency data. addition, information can guide generate pseudo labels improved semi-supervised...
Most recent scribble-supervised segmentation methods commonly adopt a CNN framework with an encoder-decoder architecture. Despite its multiple benefits, this generally can only capture small-range feature dependency for the convolutional layer local receptive field, which makes it difficult to learn global shape information from limited provided by scribble annotations. To address issue, paper proposes new CNN-Transformer hybrid solution medical image called ScribFormer. The proposed...
Glucose and lipid metabolism disorder in diabetes mellitus often causes damage to multiple tissues organs. Diabetes is beneficially affected by quercetin. However, its concrete mechanisms are yet be fully elucidated. In our study, was induced Sprague-Dawley rats STZ injection. The were randomly divided into normal control, diabetic model, low-dose quercetin treatment, high-dose pioglitazone treatment groups. Fasting blood glucose collected evaluate diabetes. Immunohistochemistry fluorometric...
In this paper, a sufficient condition for quadric surface to be an ellipsoid has been developed and closed-form solution fitting is based on constraint, which best fit the given data amongst those ellipsoids whose short radii are at least half of their major radii, in sense algebraic distance. A simple search procedure proposed pursuit 'best' when cannot well described by type ellipsoid. The algorithm quick, stable insensitive small errors data.
Detecting pneumonia, especially coronavirus disease 2019 (COVID-19), from chest X-ray (CXR) images is one of the most effective ways for diagnosis and patient triage. The application deep neural networks (DNNs) CXR image classification limited due to small sample size well-curated data. To tackle this problem, article proposes a distance transformation-based forest framework with hybrid-feature fusion (DTDF-HFF) accurate classification. In our proposed method, hybrid features are extracted...
The orthogonal design method was used to determine the optimum conditions for modifying poplar fibers through a high temperature and pressurized steam treatment subsequent preparation of wood fiber/high-density polyethylene (HDPE) composites. extreme difference, variance, significance analyses were performed reveal effect modification parameters on mechanical properties prepared composites, they yielded consistent results. main findings indicated that most strongly affected followed by...
2D splines are a powerful tool for shape modeling, either parametrically or implicitly. However, compared with regular grid-based tensor-product splines, most of the high-dimensional spline techniques based on nonregular polygons, such as box and simplex spline, generally very expensive to evaluate. Though they have many desirable mathematical properties been proved theoretically be in graphics not convenient modeling technique terms practical implementation. In practice, we still lack...
Fingerprint orientation field (FOF) estimation plays a key role in enhancing the performance of automated fingerprint identification system (AFIS): accurate FOF can evidently improve AFIS. However, despite enormous attention on research past decades, FOFs, especially for poor-quality fingerprints, still remains challenging task. In this paper, we devote to review and categorization large number methods proposed specialized literature, with particular most recent work area. Broadly speaking,...
Indoor object recognition is a key task for indoor navigation by mobile robots. Although previous work has produced impressive results in recognizing known and familiar objects, the research of robot still insufficient. In order to improve detection precision, our study proposed prior knowledge-based deep learning method aimed enable recognize objects on sight. First, we integrate public dataset private frames videos (FoVs) train convolutional neural network (CNN). Second, mean images, which...
Deep learning has been used in many computer-vision-based industrial Internet of Things applications. However, deep neural networks are vulnerable to adversarial examples that have crafted specifically fool a system while being imperceptible humans. In this article, we propose consensus defense (Cons-Def) method defend against attacks. Cons-Def implements classification and detection based on the classifications augmented examples, which generated an individually implemented intensity...
Segmentation of COVID-19 lesions can assist physicians in better diagnosis and treatment COVID-19. However, there are few relevant studies due to the lack detailed information high-quality annotation dataset. To solve above problem, we propose C2FVL, a Coarse-to-Fine segmentation framework via Vision-Language alignment merge text containing number specific locations image information. Introducing allows network achieve prediction results on challenging datasets. We conduct extensive...
Volume rendering has long been used as a key technique for volume data visualization, which works by using transfer function to map color and opacity each voxel. Many approaches proposed so far voxels classification have limited in single global function, is general unable properly visualize interested structures. In this paper, we propose localized visualization approach regards combination of two mutually related processes: the segmentation structures locally designed individual structure...
Vasculature geometry reconstruction from volumetric medical data is a crucial task in the development of computer guided minimally invasive vascular surgery systems. In this paper, technique for reconstructing vasculatures using bivariate implicit splines developed. With proposed technique, an representation tree can be accurately constructed based on voxels extracted directly surface certain structure given dataset. Experimental results show that geometric built our method faithfully...
Intensity inhomogeneity occurs in many medical images, especially vessel images. Overcoming the difficulty due to image is crucial for segmentation of image. This paper proposes a localized hybrid level-set method 3D The proposed integrates both local region information and boundary segmentation, which essential accurate extraction tiny structures. intensity firstly embedded into region-based contour model, then incorporated formulation geodesic active model. Compared with preset global...
Indoor object recognition is a key task for mobile robot indoor navigation. In this paper, we proposed pipeline detection based on convolutional neural network (CNN). With the method, first pre-train an off-line CNN model by using both public Dataset and private frames of videos (FoV) dataset. This then followed selective search process to extract region interest (RoI) after input video was parsed into frame images. The extracted RoIs are classified candidates pre-trained deep between...