A5100992291- Domain Adaptation and Few-Shot Learning
- Multimodal Machine Learning Applications
- Spectroscopy and Chemometric Analyses
- Tactile and Sensory Interactions
- Machine Learning and ELM
- Industrial Vision Systems and Defect Detection
- Handwritten Text Recognition Techniques
- Mechanical Failure Analysis and Simulation
- Material Selection and Properties
- Engineering Structural Analysis Methods
- Remote Sensing and Land Use
- Spectroscopy Techniques in Biomedical and Chemical Research
- Advanced Chemical Sensor Technologies
- Agricultural Engineering and Mechanization
- Remote-Sensing Image Classification
- Image Retrieval and Classification Techniques
- COVID-19 diagnosis using AI
- Advanced Image and Video Retrieval Techniques
- Music and Audio Processing
- Coal and Coke Industries Research
- Granular flow and fluidized beds
- Human Pose and Action Recognition
- Soil Mechanics and Vehicle Dynamics
- Mineral Processing and Grinding
China University of Mining and Technology
2024-2025
Soochow University
2024
Fudan University
2023
Xiamen University
2023
Purple Mountain Laboratories
2023
University of Windsor
2017
Few-shot image classification aims to classify images from unseen novel classes with few samples. Recent works demonstrate that deep local descriptors exhibit enhanced representational capabilities compared image-level features. However, most existing methods solely rely on either employing all or directly utilizing partial descriptors, potentially resulting in the loss of crucial information. Moreover, these primarily emphasize selection query while overlooking support descriptors. In this...
Aimed at the limitation that existing hyperspectral classification methods were mainly oriented to small-scale images, this paper proposed a new large-scale remote sensing image method, LS3EU-Net++ (Lightweight Encoder and Integrated Spatial Spectral Squeeze Excitation U-Net++). The method optimized U-Net++ architecture by introducing lightweight encoder combining (S3E) Attention Module, which maintained powerful feature extraction capability while significantly reducing training cost. In...
Proximate analysis, including ash, volatile matter, moisture, fixed carbon, and calorific value, is a fundamental aspect of fuel testing serves as the primary method for evaluating coal quality, which critical processing utilization coal. The traditional analytical methods involve time-consuming costly combustion processes, particularly when applied to large volumes that need be sampled in massive batches. Hyperspectral imaging promising rapid nondestructive determination quality indices. In...
Silos are structures made of steel commonly used as storage facilities for grains and other bulk foods. This study presents monitoring the structural behavior a recently constructed large field silo structure subjected to static grain load. Challenging full-scale tests were conducted on that measured 14.55 m in diameter 23.27 height. It was located open farm land near Bothwell, Ontario, Canada. The silo's internal pressure from loading its wall's displacement strain measured. A complex...
The texture recognition can provide clues for robots to interact with the external environment. traditional tactile material task is studied under close-set assumption, which means that all types of materials are included in training set. However, open-set much greater significance because real-world applications, there usually something doesn't belong any known class. Up now, no researcher further discussion this problem. To cope unknown classes, study proposes Open set Material Recognition...