A5100664805- Generative Adversarial Networks and Image Synthesis
- Digital Media Forensic Detection
- Advanced Image Processing Techniques
- Image Enhancement Techniques
- Advanced Vision and Imaging
- Underwater Vehicles and Communication Systems
- Insect Resistance and Genetics
- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Studies on Chitinases and Chitosanases
- Underwater Acoustics Research
- Advanced Image and Video Retrieval Techniques
- Computer Graphics and Visualization Techniques
- Nematode management and characterization studies
- Domain Adaptation and Few-Shot Learning
- Topic Modeling
- Landslides and related hazards
- Genomics and Phylogenetic Studies
- Coral and Marine Ecosystems Studies
- Cancer-related molecular mechanisms research
- Robotics and Sensor-Based Localization
- Identification and Quantification in Food
- Geophysical Methods and Applications
- Maritime Navigation and Safety
- Infrared Target Detection Methodologies
Hong Kong University of Science and Technology
2021-2024
University of Hong Kong
2021-2024
Wuhan Polytechnic University
2022-2024
Shandong University
2023-2024
Hubei Academy of Agricultural Sciences
2024
Institute of Plant Protection
2024
Sheng Jing Hospital
2024
Zhejiang University
2024
Ocean University of China
2017-2023
Zhejiang University of Technology
2023
Abstract ImageGP is an extensively utilized, open‐access platform for online data visualization and analysis. Over the past 7 years, it has catered to more than 700,000 usages globally, garnering substantial user feedback. The updated version, 2 (available at https://www.bic.ac.cn/BIC ), introduces a redesigned interface leveraging cutting‐edge web technologies enhance functionality interaction. Key enhancements include following: (i) Addition of modules format transformation, facilitating...
Abstract CRISPR-Cas systems are considered as barriers to horizontal gene transfer (HGT). However, the influence of such on HGT within species is unclear. Also, little known about impact bacterial evolution at population level. Here, using Bacillus cereus sensu lato model, we investigate interplay between and scale. We found that only a small fraction strains have (13.9% 1871), most defective based their content analysis. Comparative genomic analysis revealed this group, since harboring...
The 1 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">st</sup> Workshop on Maritime Computer Vision (MaCVi) 2023 focused maritime computer vision for Unmanned Aerial Vehicles (UAV) and Surface Vehicle (USV), organized several subchallenges in this domain: (i) UAV-based Object Detection, (ii) Mar-itime Tracking, (iii) USV-based Obstacle Segmentation (iv) Detection. were based the SeaDronesSee MODS benchmarks. This report summarizes main findings...
Abstract In the soil environment, Bacilli can affect nematode development, fecundity and survival. However, although many Bacillus species kill nematodes, virulence mechanisms utilize remain unknown. this study, we collected 120 strains comprising 30 across Bacillaceae Paenibacillaceae families of Bacillales order measured their nematicidal activities in vitro . Comparison these strains’ capacities revealed that nine species, including thuringiensis, B. cereus, subtilis, pumilus, firmus,...
Underwater image datasets are crucial in underwater vision research. Because of the strong absorption and scattering effects that occur underwater, some ground truth such as depth map, which can be easily collected in-air, becomes a great challenge environments. To solve issues associated with lack truth, we propose trainable end-to-end system an multistyle generative adversarial network (UMGAN) takes advantage cycle-consistent (CycleGAN) conditional networks. This generate multiple...
Underwater monocular visual simultaneous localization and mapping (SLAM) plays a vital role in underwater computer vision robotic perception fields. Unlike the autonomous driving or aerial environment, performing robust accurate SLAM is tough challenging due to complex aquatic environment collected critically degraded image quality. The images’ poor visibility, low contrast, color distortion result ineffective insufficient feature matching, leading even failure of existing algorithms. To...
Optical-Wireless (FiWi) access network is a newly emerged architecture which integrates passive optical networks (PONs) with wireless mesh (WMNs) to provide the ubiquitous, low cost, high bandwidth last mile Internet access. Though PON subnetwork of FiWi can bandwidth, interference in still limits throughput if all traffic goes online Internet. However, when peer-to-peer communication from one client another introduced, proposed integration PONs and WMNs significantly improve throughput. In...
Background: Infectious mononucleosis (IM) is an infectious disease with different stages of onset, and the pathogenesis IM remains unclear. Objective: To investigate levels IL-10 family cytokines IL-10, IL-22, IL-24 IL-26 in plasma samples patients (IM), potential diagnostic values. Methods: Thirty children were enrolled study collected, 30 healthy served as controls. Plasma from controls expression IL-24, IL-26, alanine aminotransferase (ALT) aspartate (AST) have been determined by Elisa...
In this paper, an energy-saving control system for intelligent street lamps based on STM32F103C8T6 microcontroller is designed, and dynamic lighting realized by integrating D80NK infrared sensor, LD2451 radar sensor KDT331 photoresistor. According to the requirements of "Urban Road Lighting Design Standard" (CJJ45-2015), adopts three operation modes: maintaining lowest illumination (5lx) when there no car, "lighting with vehicle" traffic flow sparse, full power dense. The hardware design...
We aim to improve the performance of Multiple Object Tracking and Segmentation (MOTS) by refinement. However, it remains challenging for refining MOTS results, which could be attributed that appearance features are not adapted target videos is also difficult find proper thresholds discriminate them. To tackle this issue, we propose a self-supervised (i.e., ReMOTS) framework. ReMOTS mainly takes four steps refine results from data association perspective. (1) Training encoder using predicted...
Late-time negative responses in central-loop transient electromagnetic (TEM) data are often linked to the induced polarization (IP) effect. Early methods for modeling IP effect TEM try avoid calculating fractional derivative arising from considering Cole-Cole model by either using Fourier transform convert frequency-domain time domain or approximating directly. The frequency-to-time conversion method suffer accuracy issues if number of frequencies calculated is small. time-domain...
It is well known that humans can learn and recognize objects effectively from several limited image samples. However, learning just a few images still tremendous challenge for existing main-stream deep neural networks. Inspired by analogical reasoning in the human mind, feasible strategy to “translate” abundant of rich source domain enrich relevant yet different target with insufficient data. To achieve this goal, we propose novel, effective multi-adversarial framework...
Nowadays, as a sub topic of computer vision and fishery industry, fish recognition is still challenging work not only because biodiversity, but also the complex background images. In this paper, we aim to classify different images obtained from cameras fishing vessels. Fish classification vessel camera with best known most well-investigated object problem. always take small part whole image. For these two reasons, it's challenge for us detect froma vessel. Our done Kaggle dataset. The...
Underwater depth prediction plays an important role in underwater vision research. Because of the complex environment, it is extremely difficult and expensive to obtain datasets with reliable annotation. Thus, map estimation a data-driven manner still challenging task. To tackle this problem, we propose end-to-end system including two different modules for image synthesis estimation, respectively. The former module aims translate hazy in-air RGB-D images multi-style realistic synthetic while...