A5014712849- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Hepatitis B Virus Studies
- Hepatitis C virus research
- Liver Disease Diagnosis and Treatment
- Visual Attention and Saliency Detection
- Image Enhancement Techniques
- Face and Expression Recognition
- Fault Detection and Control Systems
- Advanced Image Processing Techniques
- Target Tracking and Data Fusion in Sensor Networks
- Advanced Image Fusion Techniques
- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Generative Adversarial Networks and Image Synthesis
- Numerical methods in engineering
- Advanced Vision and Imaging
- Video Analysis and Summarization
- Multimodal Machine Learning Applications
- Text and Document Classification Technologies
- Face recognition and analysis
- Face Recognition and Perception
- Infrared Target Detection Methodologies
- Recommender Systems and Techniques
Dalian Minzu University
2019-2025
Center for High Pressure Science & Technology Advanced Research
2025
Sinovac Biotech
2025
Berry Oncology (China)
2023
Beijing Institute of Technology
2018-2021
Peking University
2021
Liaoning University of Technology
2010-2019
Huazhong University of Science and Technology
2018-2019
Beihang University
2013
University of Science and Technology Beijing
2006-2010
Salient object detection (SOD) is an important preprocessing operation for various computer vision tasks. Most of existing RGB-D SOD models employ additive or connected strategies to directly aggregate and decode multi-scale features predict salient maps. However, due the large differences between different scales, these aggregation adopted may lead information loss redundancy, few methods explicitly consider how establish connections at scales in decoding process, which consequently...
Camouflaged Object Detection (COD) is a challenging visual task due to its complex contour, diverse scales, and high similarity the background. Existing COD methods encounter two predicaments: One that they are prone falling into local perception, resulting in inaccurate object localization; Another issue difficulty achieving precise segmentation lack of detailed information. In addition, most typically require larger parameter amounts higher computational complexity pursuit better...
Multi-behavior recommendation exploits multiple types of user-item interactions, such as view and cart , to learn user preferences has demonstrated be an effective solution alleviate the data sparsity problem faced by traditional models that often utilize only one type interaction for recommendation. In real scenarios, users take a sequence actions interact with item, in order get more information about item thus accurately evaluate whether fits their personal preferences. Those behaviors...
The goal of multilabel classification is to reveal the underlying label correlations boost accuracy tasks. Most existing classifiers attempt exhaustively explore dependency between correlated labels. It increases risk involving unnecessary dependencies, which are detrimental performance. Actually, not all indispensable model. Negligible or fragile cannot be generalized well testing data, especially if there exists correlation discrepancy training and sets. To minimize such negative effect in...
The past decade has witnessed great progress in RGB-D salient object detection (SOD). However, there are two bottlenecks that limit its further development. first one is low-quality depth maps. Most existing methods directly use raw maps to perform detection, but images can bring negative impacts the performance. Hence, it not desirable utilize indiscriminately. other how effectively predict with clear boundary and complete region. To address these problems, an Attention-Guided...
Deep convolutional neural network (DCNN) is a powerful method of learning image features with more discriminative and has been studied deeply applied widely in the field computer vision pattern recognition. In order to further explore superior performance DCNN improve accuracy scene classification, this paper presents novel algorithm which fully deep characteristics images based on classical Alex-Net model support vector machine. first place, we use extract last layer 4096 neurons as method;...
This paper presents a novel multiscale neighborhood normalization-based multiple dynamic principal component analysis (MNN-MDPCA) method to detect the fault in complex batch processes with frequent operations. Since difference between batches is larger under random operations according phase, corresponding monitoring model should be changed accordingly. However, data quantity small single operation at each similar can clustered together. Due operations, follows non-Gaussian distribution. A...