A5030548422- Industrial Vision Systems and Defect Detection
- Text and Document Classification Technologies
- Smart Agriculture and AI
- Thermal Radiation and Cooling Technologies
- Imbalanced Data Classification Techniques
- Remote Sensing in Agriculture
- Hand Gesture Recognition Systems
- Natural Language Processing Techniques
- Muscle activation and electromyography studies
- Gait Recognition and Analysis
- Machine Fault Diagnosis Techniques
- Domain Adaptation and Few-Shot Learning
- Advanced Thermodynamics and Statistical Mechanics
- Speech and Audio Processing
- EEG and Brain-Computer Interfaces
- Blind Source Separation Techniques
- Topic Modeling
Chinese Academy of Sciences
2022-2024
Hefei Institutes of Physical Science
2024
University of Chinese Academy of Sciences
2022-2023
Institute of Information Engineering
2023
Changchun Institute of Optics, Fine Mechanics and Physics
2023
Institute of Acoustics
2022
Strawberry maturity detection plays an essential role in modern strawberry yield estimation and robot-assisted picking sorting. Due to the small size complex growth environment of strawberries, there are still problems with existing recognition systems’ accuracy classifications. This article proposes a algorithm based on improved YOLOv5s model named YOLOv5s-BiCE. is replacement upsampling CARAFE module structure. It improvement previous terms its content-aware processing; it also widens...
An innovative class subdomain adaptation network (CSAN) is proposed in response to the problem that distribution of bearing data inconsistent under variable working conditions, and trained on from source condition cannot be applied target condition. The CSAN model divided into two components. first component a lightweight channel convolution neural (CCNN) designed fully utilize information perform feature extraction classification. second an domain algorithm can used as loss function...
Few-shot Named Entity Recognition (NER) aims to recognize unseen name entities based on a tiny support set that consists of seen and labels, which is obviously different from traditional supervised NER methods. Contrastive learning has become popular solution for few-shot NER, improves the robustness handle unlabeled by similarity metric measure semantic between test samples entity labels. However, existing contrastive methods individually learn word embedding in source target domains,...
Thermal radiation represents a ubiquitous aspect of nature. Any object at finite temperatures emits thermal due to thermally induced motion particles and quasiparticles. The sun 6000 K the cold universe 3 K, representing most important renewable energy resource ultimate thermodynamic heat sink, are all accessible through radiation. ability control therefore plays fundamentally role in applications. Nanophotonic structures, where least one structural features wavelength or sub-wavelength...