A5103079800- Face recognition and analysis
- Advanced Image and Video Retrieval Techniques
- Generative Adversarial Networks and Image Synthesis
- 3D Shape Modeling and Analysis
- Music and Audio Processing
- Video Analysis and Summarization
- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Human Pose and Action Recognition
- Face and Expression Recognition
- Emotion and Mood Recognition
- Visual Attention and Saliency Detection
- Advanced Image Processing Techniques
- Energy Load and Power Forecasting
- Facial Rejuvenation and Surgery Techniques
- Power Transformer Diagnostics and Insulation
- Educational Reforms and Innovations
- Machine Fault Diagnosis Techniques
- Integrated Energy Systems Optimization
- Dermatologic Treatments and Research
- Gaze Tracking and Assistive Technology
- Gear and Bearing Dynamics Analysis
- Single-cell and spatial transcriptomics
- Multimodal Machine Learning Applications
- Advanced Image Fusion Techniques
NetEase (China)
2020-2023
Yanshan University
2022
Zhejiang University
2015-2021
Harbin Institute of Technology
2021
China University of Petroleum, East China
2020
Cell types are the basic building units of multicellular life, with extensive diversities. The evolution cell is a crucial layer comparative biology but thus far not comprehensively studied. We define compendium atlases using single-cell RNA-seq (scRNA-seq) data from seven animal species and construct cross-species cell-type evolutionary hierarchy. present roadmap for origin diversity major categories find that muscle neuron cells conserved types. Furthermore, we identify transcription...
How to manage, store, and index large numbers of videos is an urgent problem be solved. Although there are many video summarization models achieving good results, based on low-level features cannot summarize important semantic information analysis need related text descriptions that do not exist for most videos. As a consequence, the mining contained in itself more feasible way. In this paper, we propose action parsing-driven model reinforcement learning. The mainly divided into two parts,...
The novel rotating synthetic aperture (RSA) optical imaging system is an important development direction for future high-resolution remote sensing satellites in geostationary orbit. However, owing to the rectangular pupil, point spread function of RSA has asymmetric spatial distribution, and images obtained using primary mirror from different rotation angles have nonuniform blur degradation. Moreover, platform vibration pupil coupling effects on imaging, resulting further radiometric...
A large number of videos are generated and uploaded to video websites (like youku, youtube) every day play more important roles in human life. While bringing convenience, the big data raise difficulty summarization allow users browse a easily. However, although there many existing approaches, key frames selected fail integrate contexts qualities summarized results difficult evaluate because lack ground-truth. Inspired by previous methods that extract frames, we propose deep recurrent neural...
In this work, we propose a stroke-based hairstyle editing network, dubbed HairstyleNet, allowing users to conveniently change the hairstyles of an image in interactive fashion. Different from previous works, simplify process where can manipulate local or entire by adjusting parameterized hair regions. Our HairstyleNet consists two stages: stroke parameterization stage and stroke-to-hair generation stage. stage, first introduce parametric strokes approximate wisps, shape is controlled...
Game character customization is one of the core features many recent Role-Playing Games (RPGs), where players can edit appearance their in-game characters with preferences. This paper studies problem automatically creating a single photo. In literature on this topic, neural networks are introduced to make game engine differentiable and self-supervised learning used predict facial parameters. However, in previous methods, expression parameters identity highly coupled each other, making it...
Visual separability between different objects in various image classification tasks is highly uneven. As a consequence, humans need levels of detailed descriptions to separate multi-granularity similarities. Meanwhile, deep networks, such as convolutional neural networks (C-NNs) have demonstrated great ability multilevel representations for an object. Unfortunately, existing methods with typically use the output last layer only feature train flat N-way classifiers, which fail fit character....
As a high-quality secondary energy, hydrogen energy has great potential in storage and utilization. The development of power-to-hydrogen (P2H) technology alleviated the problem wind curtailment improved coupling between power grid natural gas grid. Under premise ensuring safety, using P2H to mix produced into network for long-distance transmission generation can not only promote but also reduce carbon emissions. This paper presents new model incorporating pipelines. To minimize sum cost,...
Abstract As the condition monitoring and control device in distribution automation system, abnormal or fault state of terminal units’ measurement system will negatively affect quality measured electrical quantities, therefore, fast accurate discrimination state’s data improve reliability system. This paper proposes a method, which is based on generative adversarial network (GAN) combined with convolutional neural (CNN), to discriminate specific category terminals’ measuring data. Firstly,...
The automatic intensity estimation of facial action units (AUs) from a single image plays vital role in analysis systems. One big challenge for data-driven AU is the lack sufficient label data. Due to fact that annotation requires strong domain expertise, it expensive construct an extensive database learn deep models. limited number labeled AUs as well identity differences and pose variations further increases difficulties. Considering all these difficulties, we propose unsupervised...
Facial action unit (AU) intensity is an index to describe all visually discernible facial movements. Most existing methods learn estimator with limited AU data, while they lack of generalization ability out the dataset. In this paper, we present a framework predict parameters (including identity and parameters) based on bone-driven face model (BDFM) under different views. The proposed consists feature extractor, generator, parameter regressor. regressor can fit physical meaning BDFM from...
Detecting saliency objects in video is a challenging problem. Conventional detection methods for still images do not take consideration of the motion information, which may fail to detect moving videos. In this paper, we propose novel method detecting Motion cues, are extracted from both image orientations and orientations, integrated with cues order find objects, We extract "compositions" each frame reform potential shape salient object. Additionally, introduce an extended Spatial-temporal...
Traditional operas account for an important part of Chinese intangible cultural heritages and are embodiments a nation's soft power.Lv opera, national heritage, drives from Dongying in Shandong province, China.The opera is simple, unadulterated, fluent easy to learn, thus popular among the local people.As time moves forward, Lv has unwittingly become indispensable culture Shandong.Yet, as most audience old people rural areas, college students largely know little about this form art.They pay...
Game character customization is one of the core features many recent Role-Playing Games (RPGs), where players can edit appearance their in-game characters with preferences. This paper studies problem automatically creating a single photo. In literature on this topic, neural networks are introduced to make game engine differentiable and self-supervised learning used predict facial parameters. However, in previous methods, expression parameters identity highly coupled each other, making it...