A5044727487- Smart Agriculture and AI
- Industrial Vision Systems and Defect Detection
- Advanced Neural Network Applications
- Spectroscopy and Chemometric Analyses
- Advanced Chemical Sensor Technologies
- Prostate Cancer Diagnosis and Treatment
- Image Processing Techniques and Applications
- Prostate Cancer Treatment and Research
- Advanced Measurement and Detection Methods
- Image and Object Detection Techniques
- Plant Pathogens and Fungal Diseases
- Renal cell carcinoma treatment
- Plant Disease Management Techniques
- Block Copolymer Self-Assembly
- Structural Engineering and Vibration Analysis
- Dendrimers and Hyperbranched Polymers
- Brain Tumor Detection and Classification
- Remote-Sensing Image Classification
- Pancreatitis Pathology and Treatment
- 3D Shape Modeling and Analysis
- Traditional Chinese Medicine Studies
- Polymer composites and self-healing
- Surface Roughness and Optical Measurements
- Vibration Control and Rheological Fluids
- Vehicle License Plate Recognition
Beijing Academy of Artificial Intelligence
2025
Zhejiang University
2005-2024
Peking University Third Hospital
2024
Ministry of Education of the People's Republic of China
2024
Peking University
2024
Anhui Agricultural University
2023
Huazhong Agricultural University
2023
Qilu University of Technology
2023
Anhui Provincial Meteorological Bureau
2023
Fourth Affiliated Hospital of Anhui Medical University
2022
The accurate prevention and control of pear tree diseases, especially the precise segmentation leaf poses a serious challenge to fruit farmers globally. Given possibility disease areas being minute with ambiguous boundaries, becomes difficult. In this study, we propose model named MFBP-UNet. It is based on UNet network architecture integrates Multi-scale Feature Extraction (MFE) module Tokenized Multilayer Perceptron (BATok-MLP) dynamic sparse attention. MFE enhances extraction detail...
The electronic nose system is widely used in tea aroma detecting, and the sensor array plays a fundamental role for obtaining good results. Here, optimization (SAO) method based on correlation coefficient cluster analysis (CA) proposed. First, distinguishing performance value (DPV) are calculated to eliminate redundant sensors. Then, independence obtained through number of sensors confirmed. Finally, optimized constructed. According results proposed method, green (LG), fried (LF) baked (LB)...
In this study, an embedded machine vision system using Gabor filters and Pulse Coupled Neural Network (PCNN) is developed to identify defects of warp-knitted fabrics automatically. The consists smart cameras a Human Machine Interface (HMI) controller. A hybrid detection algorithm combing PCNN running on the SOC processor camera. First, are employed enhance contrast images captured by CMOS sensor. Second, defect areas segmented with adaptive parameter setting. Third, will notice controller...
Abstract For practitioners, it is very crucial to realize accurate and automatic vision-based quality identification of Longjing tea. Due the high similarity between classes, classification accuracy traditional image processing combined with machine learning algorithm not satisfactory. High-performance deep methods require large amounts annotated data, but collecting labeling massive data time consuming monotonous. To gain as much useful knowledge possible from related tasks, an...
Car fine recognition is a typical scenario for fine-grained image classification, which has great research and application value in both civilian military fields. However, current on classification often limited to improving the accuracy of models, ignoring need lightweight efficient applications practical applications, resulting disconnect from reality. In this paper, car method based attention network regularized fine-tuning proposed. Based high-performance, convolutional neural (CNN)...
In the fabric manufacturing industry, defect detection is a practical yet challenging task, due to problem of defects with small sizes or unremarkable appearances distributed in images high resolution. Some deep‐learning‐based solutions try tackle aforementioned but limited achievements. Herein, brand new module called adaptively fused attention (AFAM) proposed improve performance by enabling network concentrate more on terms 1) enhancing feature maps both spatial‐wise and channel‐wise, 2)...
Image datasets acquired from orchards are commonly characterized by intricate backgrounds and an imbalanced distribution of disease categories, resulting in suboptimal recognition outcomes when attempting to identify apple leaf diseases. In this regard, we propose a novel model, named RFCA ResNet, equipped with dual attention mechanism multi-scale feature extraction capacity, more effectively tackle these issues. The incorporated into ResNet is potent tool for mitigating the detrimental...
Abstract Image-based fruit classification offers many useful applications in industrial production and daily life, such as self-checkout the supermarket, automatic sorting dietary guidance. However, task will have different data distributions due to application scenarios. One feasible solution solve this problem is use domain adaptation that adapts knowledge from original training (source domain) new testing (target domain). In paper, we propose a novel deep learning-based unsupervised...
Abstract Different quality grades of tea tend to have a high degree similarity in appearance. Traditional image‐based identification methods limited effects, while complex deep learning architectures require much data and long‐term training. In this paper, two based on convolutional neural networks transfer are proposed. types images collected by self‐designed computer vision system form set, which is small‐scale inter‐ intraclass similarity. The first method uses three simplified network...
The developing of satellite remote sensing technology demands high precision reflectors. This paper is devoted to the development an ultra-high reflector, including structural design and optimization, even a kind zero-expansion composite. A novel parabolic reflector without traditional back surface designed, simulated analyzed. relationships some key parameters(surface error weight) with other parameters are obtained. As Kevlar fabric has negative CTE (Coefficient Thermal Expansion) CFRP...
The traditional defect detection algorithms based on image registration, contrast and other processing are only limited to a single defect. Though deep-learning-based object can be used detect variety of different defects, the state-of-the-art still have low accuracy small size defects. Basing Cascade R-CNN in this paper, new multi-scale feature extraction method—the Multi-Scale Feature Pair—is proposed is establish model for metal products an enterprise. Experimental results show that...
The Revised Atlanta Classification (RAC) and Determinant-Based (DBC) are currently two widely adopted systems for evaluating the severity of acute pancreatitis (AP). This study aimed to overcome inaccuracies limitations that existed in them.We retrospectively analyzed 298 patients with AP. "Two-Step" approach was divided into an early organ failure (OF) assessment: (I) none, (II) transient, (III) single persistent, (IV) multiple persistent; a later local complications (A) (B) sterile, (C)...
Abstract In agricultural production, pest problems are inevitable. recent years, China has suffered annual losses of up to 40 million tons grain due various pests and diseases. There over one known insect species in the natural world, exhibiting complex diverse morphologies, making manual identification costly. With advancement deep learning technologies, methods relying solely on image data for crop have achieved some success. However, these heavily depend numerous high-quality annotated...
Currently, the association between prostate volume (PV) or weight with pathological outcomes in patients cancer (PCa) is not well understood. This study aimed to explore whether PV can predict adverse of PCa after radical prostatectomy (RP). A total 1063 men confirmed localized who underwent RP at First Affiliated Hospital Zhejiang University from January 2014 April 2019 were retrospectively analyzed. Patients assigned into small, medium and large groups based on PV. The analysis variance,...
The next generation wireless LAN standard IEEE 802.1 In adopts MIMO-OFDM (multi-input multi-output orthogonal frequency division multiplexing) as a key technology in physical layer. this paper, the effect of sampling clock offset is analyzed, an algorithm estimation proposed on system presence offset. based long training field preamble, utilizing correlation between two same symbols. It can perform synchronization efficiently and quickly. simulation results show that when SNR...