A5101802990- Face and Expression Recognition
- Image Retrieval and Classification Techniques
- Traditional Chinese Medicine Studies
- Neural Networks and Applications
- Emotion and Mood Recognition
- Advanced Image and Video Retrieval Techniques
- Domain Adaptation and Few-Shot Learning
- Machine Learning and Data Classification
- Face recognition and analysis
- Advanced Neural Network Applications
- Gene expression and cancer classification
- Text and Document Classification Technologies
- Multimodal Machine Learning Applications
- Imbalanced Data Classification Techniques
- Video Surveillance and Tracking Methods
- COVID-19 diagnosis using AI
- Evolutionary Algorithms and Applications
- Advanced Computing and Algorithms
- Cancer-related molecular mechanisms research
- Speech and Audio Processing
- Topic Modeling
- Anomaly Detection Techniques and Applications
- Natural Language Processing Techniques
- Remote-Sensing Image Classification
- Music and Audio Processing
South China University of Technology
2016-2025
Hubei University
2023
Songshan Lake Materials Laboratory
2023
Southern Medical University
2023
Guangzhou University
2023
Hubei University for Nationalities
2023
South China Botanical Garden
2021
University of Chinese Academy of Sciences
2021
Beijing Botanical Garden
2021
National University of Defense Technology
2018
Facial expression recognition (FER) becomes more challenging in the wild due to unconstrained conditions, such as different illumination, pose changes, and occlusion of face. Current FER methods deploy attention mechanism deep neural networks improve performance. However, these models only capture limited features relationships. Thus this paper proposes a novel framework called multi-relations aware network (MRAN), which can focus on global local learn multi-level relationships among...
Facial expression recognition has been widely used to solve the problems such as lie detection and human-machine interaction. However, due difficulties control application environments, current methods have lower accuracy in practice. This paper proposes a new method for facial by considering several aspects. First, human beings are easy recognize some expressions, while difficult others. Inspired this intuition, loss function is proposed enlarge distances between samples from easily...
Medical images such as facial and tongue have been widely used for intelligence-assisted diagnosis, which can be regarded the multi-label classification task disease location (DL) nature (DN) of biomedical images. Compared with complicated convolutional neural networks Transformers this task, recent MLP-like architectures are not only simple less computationally expensive, but also stronger generalization capabilities. However, models require better input features from image. Thus, study...
Facial expression recognition (FER) has a wide range of applications, including interactive gaming, healthcare, security, and human‐computer interaction systems. Despite the impressive performance FER based on deep learning methods, it remains challenging in real‐world scenarios due to uncontrolled factors such as varying lighting conditions, face occlusion, pose variations. In contrast, humans are able categorize objects both their inherent characteristics surrounding environment from...
Residual networks, which use a residual unit to supplement the identity mappings, enable very deep convolutional architecture operate well, however, has been proved be diverse and redundant, may leads low-efficient modeling. In this work, we propose competitive squeeze-excitation (SE) mechanism for network. Re-scaling value each channel in structure will determined by mappings jointly, design enables us expand meaning of relationship modeling blocks. Modeling competition between cause flow...
The tongue image provides important physical information of humans. It is great importance for diagnoses and treatments in clinical medicine. Herbal prescriptions are simple, noninvasive, have low side effects. Thus, they widely applied China. Studies on the automatic construction technology herbal based images significance deep learning to explore relevance prescriptions, it can be healthcare services mobile medical systems. In order adapt a variety photographic environments construct...
Food recognition plays a much critical role in various health-care applications. However, it poses many challenges to current approaches due the diverse appearances of food dishes and non-uniform composition ingredients for foods same category. Current methods primarily focus on appearance without considering their semantic information, easily finding wrong attention areas images. Second, these lack dynamic weighting multiple features modeling process. Thus this paper proposes novel...
Supervised machine learning has several drawbacks that make it difficult to use in many situations. Drawbacks include heavy reliance on massive training data, limited generalizability, and poor expressiveness of high-level semantics. Low-shot Learning attempts address these drawbacks. allows the model obtain good predictive power with very little or no where structured knowledge plays a key role as semantic representation human. This article will review fundamental factors low-shot...
Body constitution classification is the basis and core content of traditional Chinese medicine research. It to extract relevant laws from complex phenomenon finally build system. Traditional identification methods have disadvantages inefficiency low accuracy, for instance, questionnaires. This paper proposed a body recognition algorithm based on deep convolutional neural network, which can classify individual types according face images. The model first uses network features image then...