A5103182093- 3D Shape Modeling and Analysis
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- Advanced Vision and Imaging
- 3D Surveying and Cultural Heritage
- Computer Graphics and Visualization Techniques
- Multimodal Machine Learning Applications
- Aluminum Alloy Microstructure Properties
- Pharmaceutical Economics and Policy
- Intellectual Property and Patents
- Thermal Radiation and Cooling Technologies
- Caching and Content Delivery
- Human Motion and Animation
- Solar Thermal and Photovoltaic Systems
- AI in cancer detection
- Medical Image Segmentation Techniques
- Image Processing and 3D Reconstruction
- High Temperature Alloys and Creep
- Domain Adaptation and Few-Shot Learning
- Multilevel Inverters and Converters
- Recommender Systems and Techniques
- Viral gastroenteritis research and epidemiology
- HIV/AIDS drug development and treatment
- Remote Sensing and LiDAR Applications
- Ubiquitin and proteasome pathways
North China University of Technology
2024-2025
Henan Polytechnic University
2025
Shanghai Jiao Tong University
2008-2024
Wuhan University
2024
Xi'an Jiaotong University
2023
Zhejiang University
2023
Vanderbilt University
2021-2022
University at Buffalo, State University of New York
2021
Central South University
2017
University of Macau
2013-2014
Hidden features in neural network usually fail to learn informative representation for 3D segmentation as supervisions are only given on output prediction, while this can be solved by omni-scale supervision intermediate layers. In paper, we bring the first method point cloud via proposed gradual Receptive Field Component Reasoning (RFCR), where target Codes (RFCCs) designed record categories within receptive fields hidden units encoder. Then, RFCCs will supervise decoder gradually infer a...
Boundary information plays a significant role in 2D image segmentation, while usually being ignored 3D point cloud segmentation where ambiguous features might be generated feature extraction, leading to misclassification the transition area between two objects. In this paper, firstly, we propose Prediction Module (BPM) predict boundary points. Based on predicted boundary, boundary-aware Geometric Encoding (GEM) is designed encode geometric and aggregate with discrimination neighborhood, so...
Cyber-physical systems are becoming part of our daily life, and a large number data generated at such an unprecedented rate that it becomes larger than ever before in social cyber-physical systems. As consequence, is highly desired to process these big so meaningful knowledge can be extracted from those vast diverse data. Based on large-scale data, using collaborative filtering recommendation methods recommend some valuable clients or products for e-commerce websites users considered as...
Hidden features in the neural networks usually fail to learn informative representation for 3D segmentation as supervisions are only given on output prediction, while this can be solved by omni-scale supervision intermediate layers. In paper, we bring first method via proposed gradual Receptive Field Component Reasoning (RFCR), where target Codes (RFCCs) is designed record categories within receptive fields hidden units encoder. Then, RFCCs will supervise decoder gradually infer a...
Synchronous space vector modulation (SSVM) has been widely applied in motor drives operated with a low switching frequency. For SSVM based closed-loop control, the phase angle of voltage command updated at each control step should equal to predefined sampling positions. The prior arts usually adopt proportional integral controller period compensation achieve current regulation and synchronization, which show slow dynamic responses. This paper proposes SSVM-based model predictive flux (MPFC)...
In this study, the phase transformation mechanism during decomposition of undercooled austenite and its effect on deformation behavior a high-strength medium Mn steel were studied. The results indicate that formation heating (α → γ) is relatively fast reaction. However, prior above martensite start (Ms) temperature (γ α) difficult due to high thermal stability. Only occurred final air-cooling stage following 120-h isothermal treatment at 360 °C (slightly Ms). growth laths was limited by...
Besides local features, global information plays an essential role in semantic segmentation, while recent works usually fail to explicitly extract the meaningful and make full use of it. In this paper, we propose a SceneEncoder module impose scene-aware guidance enhance effect information. The predicts scene descriptor, which learns represent categories objects existing directly guides point-level segmentation through filtering out not belonging scene. Additionally, alleviate noise region,...
Cloud computing has emerged as today’s most exciting paradigm for providing services using a shared framework, which opens new door solving the problems of explosive growth digital resource demands and their corresponding convenience. With exponential number data types size in so-called big work, backbone network is under great pressure due to its transmission capacity, lower than would seriously hinder development without an effective approach solve this problem. In paper, Intelligent...
Part parsing is taken as a dense prediction task, assigning each pixel semantic part label. Some previous methods tried to model the human-known relationships among different parts (inter-part). However, these are hard be used for multi-object since given highly dependent on human priors which require special learn. In addition, pixels in same (intra-part) always assumed equally important. fact, even they belong part, some quite uncertain their predictions while with high confidence, but...
Prior plays an important role in providing the plausible constraint on human motion. Previous works design motion priors following a variety of paradigms under different circumstances, leading to lack versatility. In this paper, we first summarize indispensable properties prior, and accordingly, framework learn versatile which models inherent probability distribution motions. Specifically, for efficient prior representation learning, propose global orientation normalization remove redundant...