A5073973915- Robotics and Sensor-Based Localization
- 3D Surveying and Cultural Heritage
- 3D Shape Modeling and Analysis
- Infrastructure Maintenance and Monitoring
- Remote Sensing and LiDAR Applications
- Glaucoma and retinal disorders
- Advanced Vision and Imaging
- Non-Destructive Testing Techniques
- Robotic Path Planning Algorithms
- High-Temperature Coating Behaviors
- Concrete Corrosion and Durability
- Advanced Neural Network Applications
- Retinal Diseases and Treatments
- High Entropy Alloys Studies
- Advanced Image and Video Retrieval Techniques
- Additive Manufacturing Materials and Processes
- Graph Theory and Algorithms
- Retinal and Macular Surgery
- Retinal and Optic Conditions
- Retinal Imaging and Analysis
- Gut microbiota and health
- Advanced Image Processing Techniques
- Genetic Associations and Epidemiology
- Domain Adaptation and Few-Shot Learning
- Autophagy in Disease and Therapy
Nanchang University
2018-2025
California Institute of Technology
2025
Hunan University
2025
Nanyang Technological University
2022-2024
Institute of Computing Technology
2024
Chinese Academy of Sciences
2024
University of Hong Kong
2024
Hong Kong University of Science and Technology
2024
University at Buffalo, State University of New York
2024
First Affiliated Hospital of Jiangxi Medical College
2024
The average dipole moment of a water molecule in the condensed phase is enhanced by around 40 percent relative to that an isolated monomer. This enhancement results from large polarization caused electric field induced surrounding monomers. A quantitative molecular description this essential for modeling aqueous solvation phenomena. combined theoretical and experimental study moments small clusters provides such also gives insights into structure liquid water.
Background Previous researches have implicated a vital association between gut microbiota (GM) and diabetic retinopathy (DR) based on the of “gut-retina” axis. But their causal relationship has not been elucidated. Methods Instrumental variables 211 GM taxa were obtained from genome wide study (GWAS), Mendelian randomization was carried out to estimate effects DR risk FinnGen GWAS (14,584 cases 202,082 controls). Inverse variance weighted (IVW) is main method analyze causality, MR results...
With the incremental quantities of variable speed wind turbines (VSWT) integrated into microgrid, frequency fluctuation and generation efficiency decrease are caused. To solve problem, power unreserved control has been well studied, which emulates inertia stimulates VSWTs to transiently release kinetic energy stored in rotating masses, thus providing response excursions promptly timely. Meanwhile, temporary over-production (TOP) control, as two main strategies have put forward improve...
This work presents FG-Net, a general deep learning framework for large-scale point cloud understanding without voxelizations, which achieves accurate and real-time performance with single NVIDIA GTX 1080 8G GPU an i7 CPU. First, novel noise outlier filtering method is designed to facilitate the subsequent high-level tasks. For effective purpose, we propose plug-and-play module consisting of correlated feature mining deformable convolution-based geometric-aware modeling, in local...
Robust point cloud parsing under all-weather conditions is crucial to level-5 autonomy in autonomous driving. However, how learn a universal 3D semantic segmentation (3DSS) model largely neglected as most existing benchmarks are dominated by clouds captured normal weather. We introduce SemanticSTF, an adverse-weather dataset that provides dense point-level annotations and allows study 3DSS various adverse weather conditions. modeling two setups: 1) domain adaptive adapts from normal-weather...
The defect diagnosis of modern infrastructures is crucial to public safety. In this work, we propose an unsupervised domain adaptive crack recognition framework. To fulfill the adaptation (UDA) task cracks in infrastructural inspections, a robust UDA learning strategy termed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Crack-DA</i> increase generalization capacity model unseen test circumstances. More specifically, first leveraging...
Purpose: To explore the mechanisms relating gut microbiome (GM) to age-related macular degeneration (AMD), as they remain unclear. GM taxa that appear act within gut–retina axis may affect risk of AMD. Methods: Single-nucleotide polymorphisms (SNPs) 196 were obtained from MiBioGen consortium, and a Mendelian randomization (MR) study was carried out estimate causality between AMD (defined an endpoint based on ICD-9 ICD-10). Using data FinnGen consortium (6157 patients 288,237 controls), we...
Automatic crack detection and segmentation plays a significant role in the whole system of unmanned aerial vehicle inspections. In this paper, we have implemented deep learning framework for based on classical network architectures including Alexnet, VGG Resnet. Moreover, inspired by feature pyramid architecture, hierarchical convolutional neural (CNN) which is efficient also proposed, performance it compared with other state art architecture. We summarized existing datasets established...
To assess the diagnostic efficacy of optical coherence tomography (OCT) and OCT angiography (OCTA) in Parkinson's disease (PD).OCT was used to obtain macular parameters peripapillary retinal nerve fiber layer (RNFL) thickness. The superficial vessel foveal avascular zone (FAZ) were quantified with OCTA. area under receiver operating characteristic curve (AUC) indicated parameters.Thirty-five eyes from 35 PD patients age-matched healthy subjects who served as controls evaluated. mean RNFL...
This work presents a general deep learning framework for large-scale point clouds understanding without voxelizations, called FG-Conv, which achieves an accurate and real-time of clouds. Through our novel design combining feature level correlation mining deformable convolutions based geometric aware modeling, the local relationships patterns can be captured. The attention mechanism is also adopted to enhance global long-range correlations. Finally, pyramid residual network proposed combine...
The current light detection and ranging simultaneous localization mapping (LiDAR-SLAM) system suffers greatly from low accuracy limited robustness when faced with complicated circumstances. From our experiments, we find that LiDAR-SLAM systems have performance specular surfaces such as glass, certain metals, building walls are rich in urban environments. Therefore, this work, propose a general termed <underline xmlns:mml="http://www.w3.org/1998/Math/MathML"...
Contrastive learning has recently demonstrated great potential for unsupervised pre-training in 3D scene understanding tasks. However, most existing work randomly selects point features as anchors while building contrast, leading to a clear bias toward background points that often dominate scenes. Also, object awareness and foreground-to-background discrimination are neglected, making contrastive less effective. To tackle these issues, we propose general foreground-aware feature contrast...
Corneal diseases present a significant challenge to global health. Given the uneven distribution of ophthalmic resources, development system facilitate remote diagnosis corneal is particularly crucial. In this study, we developed an artificial intelligence named Gancor, based on large-scale clinical dataset comprising 9669 anterior segment (AS) images and corresponding fluorescein staining (CFS) from Affiliated Eye Hospital Nanchang University, as well 967 pairs AS-CFS captured via...
Background: Early diagnosis of proliferative vitreoretinopathy (PVR) is crucial for preventing vision loss and ensuring effective treatment retinal detachment. This study developed validated a system named Artificial Intelligence-Proliferative Vitreoretinopathy (AI-PVR) Insight, which aims to automate the identification, grading, postoperative risk assessment PVR. Methods: We retrospectively collected data from 1,700 eyes patients who underwent vitrectomy at Eye Hospital affiliated with...
This work presents FG-Net, a general deep learning framework for large-scale point clouds understanding without voxelizations, which achieves accurate and real-time performance with single NVIDIA GTX 1080 GPU. First, novel noise outlier filtering method is designed to facilitate subsequent high-level tasks. For effective purpose, we propose convolutional neural network leveraging correlated feature mining deformable convolution based geometric-aware modelling, in the local relationships...
Retinoblastoma (RB) is the most common intraocular tumor among children. Leucine-rich pentatricopeptide repeat (PPR)-motif-containing protein (LRPPRC), a suppressor gene of autophagy, has been proven to play regulatory role in progression. However, little known about functional roles and mechanisms LRPPRC RB First, we performed detailed analysis for normal control. The expression tissues was significantly higher than that tissues. Moreover, suppression could repress cell migration, invasion,...