A5011916547- Advanced Neural Network Applications
- Advanced Image Processing Techniques
- Infrared Target Detection Methodologies
- Brain Tumor Detection and Classification
- Medical Image Segmentation Techniques
- Image Processing Techniques and Applications
- Retinal Imaging and Analysis
- AI in cancer detection
- Glaucoma and retinal disorders
- Advanced Measurement and Detection Methods
- Advanced Image Fusion Techniques
- Advanced Vision and Imaging
- Advanced SAR Imaging Techniques
- Image Enhancement Techniques
- Image Retrieval and Classification Techniques
- Domain Adaptation and Few-Shot Learning
- COVID-19 diagnosis using AI
- Retinal and Optic Conditions
- Space Satellite Systems and Control
- Lipid metabolism and disorders
- Optical Systems and Laser Technology
- Remote-Sensing Image Classification
- Pharmacological Effects of Natural Compounds
- Spectroscopy Techniques in Biomedical and Chemical Research
- Molecular Biology Techniques and Applications
Beihang University
2020-2025
Hefei Institute of Technology Innovation
2022-2023
China Academy of Space Technology
2021
Shandong Provincial Hospital
2020
Shandong University
2020
Shanghai Jiao Tong University
2019
Chinese Academy of Sciences
2018
Existing methods for remote sensing image dehazing and thin cloud removal treat this restoration task as a clear pixel estimation problem, yielding single prediction result through deterministic pipeline. However, is highly ill-posed the sharp value corresponding to input cannot be uniquely determined solely from degraded image. In paper, we present novel algorithm haze using Conditional Variational Autoencoders (CVAE) generate multiple realistic restored images each input. By sampling...
Excessive fructose (FRU) intake can result in insulin resistance and metabolic disorder, which are related to renal injury.18α-Glycyrrhetinic acid (GA) is a bioactive component mainly extracted from Glycyrrhiza radix, has anti-oxidant anti-inflammatory activities. However, its effects on FRU-induced injury still remain unclear. In this study, we found that 18α-GA treatments could significantly ameliorate the cell viability FRU-treated tubule epithelial cells, accompanied with improved...
Abstract Countries are increasingly interested in spacecraft surveillance and recognition which play an important role on-orbit maintenance, space docking, other applications. Traditional detection methods, including radar, have many restrictions, such as excessive costs energy supply problems. For servicing spacecraft, image is a simple but relatively accurate method for obtaining sufficient position direction information to offer services. However, the best of our knowledge, few practical...
Robust detection of infrared small target under complex background is great significance for search and tracking applications. However, the inherent problem limited prior features has always made its task a challenging research topic. In order to solve problem, we propose novel method based on monogenic signal decomposition feature expansion, which can effectively enrich extract potential target. First, series local information original image obtained through constructed by Riesz transform....
Synthetic Aperture Radar (SAR) has received more attention due to its complementary superiority on capturing significant information in the remote sensing area. However, for an Aerial View Object Classification (AVOC) task, SAR images still suffer from long-tailed distribution of aerial view objects. This disparity limit performance classification methods, especially data-sensitive deep learning models. In this paper, we propose a two-stage shake-shake network tackle problem. Specifically,...
Deep learning has been successfully applied to single image super-resolution problem due its high data fitting ability. However, the trending of deeper layers and wider receptive field acquire better performance brings computation complexity serious information vanishing. To address this problem, we proposed a new Reconstructed DenseNets model for super-resolution. The basic idea behind is improve recent by modifying two core modules, dense blocks transition blocks, so that can emphasize...
The quality of a blastocyst directly determines the embryo's implantation potential, thus making it essential to objectively and accurately identify morphology. In this work, we propose an automatic framework named I2CNet perform segmentation task in human embryo images. contains two components: IntrA-Class Context Module (IACCM) InteR-Class (IRCCM). IACCM aggregates representations specific areas sharing same category for each pixel, where categorized regions are learned under supervision...
The segmentation of blastocyst components is vital in assessing embryo quality, as implantation potential closely relates to morphological characteristics. Despite this significance, automated encounters challenges like poor contrast, noise, and indistinct boundaries among organizational structures. In response, we present a novel transformer architecture called BTFormer for segmentation. Our approach integrates an axial-free attention mechanism with reduced computational demands,...
Accurate detection and segmentation of obstacles is the key to smooth operation planetary rovers basic guarantee scientific exploration mission. The traditional method rock based on boundary detector affected by change illumination dust storms. To address this problem, paper proposes an improved U-net-based architecture combined with Visual Geometry Group (VGG) dilated convolutional neural network for rocks from images rovers. proposed also has a contracting path expansive get...
It is of paramount importance for a rover running on an extraterrestrial body surface to recognize the dangerous zones autonomously. This automation inevitable due communication delay. However, as far we know, there are few annotated terrain recognition datasets bodies. Furthermore, lack hinders training and evaluation algorithms. Therefore, first built Chang’e 3 (CE3TR) dataset address semantic segmentation problems lunar surface. The moon one nearest celestial bodies earth; our work geared...
In this paper we propose a novel model towards multi-frame super-resolution, which leverages multiple RAW images and yields super-resolved RGB image. To facilitate the pixel misalignment in burst photography, apply refined Pyramid Cascading Deformable Convolution (PCD) feature alignment module. A new 3D deformable convolution fusion module is proposed subsequently to merge information from all frames adaptively. addition, employ an encoder-decoder network restore color details sRGB space...
Accurate segmentation of the acquired 3D stack fluorescence images mice gland is a prerequisite for qualitative and quantitative analysis. However, traditional methods cannot fully satisfy need accurate reconstruction result generally low signal-to-noise ratio unique structure gland. In this article, task regarded as multi-label classification problem design to bind widely-used method U-Net up with an ensemble learning The proposed tested on our confocal gastric experiment results show that...