A5029799469- Face recognition and analysis
- ZnO doping and properties
- Face and Expression Recognition
- Gas Sensing Nanomaterials and Sensors
- Ga2O3 and related materials
- Video Surveillance and Tracking Methods
- Gait Recognition and Analysis
- Advanced Vision and Imaging
- Anomaly Detection Techniques and Applications
- Topic Modeling
- Sentiment Analysis and Opinion Mining
- Advanced Image Processing Techniques
- Thin-Film Transistor Technologies
- Natural Language Processing Techniques
- Human Pose and Action Recognition
- Nanowire Synthesis and Applications
Beijing University of Posts and Telecommunications
2023
Liaoning Normal University
2011
Self-supervised pre-training paradigms have been extensively explored in the field of skeleton-based action recognition. In particular, methods based on masked prediction pushed performance to a new height. However, these take low-level features, such as raw joint coordinates or temporal motion, targets for regions, which is suboptimal. this paper, we show that using high-level contextualized features can achieve superior performance. Specifically, propose Skeleton2vec, simple and efficient...
Arsenic-doped petal-like zinc oxide microrods are grown on silicon (100) substrates by the chemical vapor deposition method without use of catalysts. Scanning electron microscopy shows that As-doped ZnO with a preferred c-axial orientation obtained, which is well in accordance x-ray diffraction analysis. The obtained have uniform lengths about 2 μm and side 100 nm. As-related acceptor emissions observed from photoluminescence spectra at temperature 11 K. binding energy estimated to be 128 meV.
With the increasing availability of consumer depth sensors, 3D face recognition (FR) has attracted more and attention. However, data acquired by these sensors are often coarse noisy, making them impractical to use directly. In this paper, we introduce an innovative Depth map denoising network (DMDNet) based on Denoising Implicit Image Function (DIIF) reduce noise enhance quality facial images for low-quality FR. After generating clean faces using DMDNet, further design a powerful called...
One-dimensional phosphorus doped ZnO nanowires and nanonails are prepared on Si substrate without employing any metal catalyst by chemical vapor deposition method. Field-emission scanning electron microscopy shows that the samples located downstream 1.5 cm away from source material of nanowire structure 1 above materials nanonail structure, growth mechanisms nanostructures with different morphologies discussed. The photoluminescence properties studied at a temperature 10 K. related acceptor...
Historically lower-level tasks such as automatic speech recognition (ASR) and speaker identification are the main focus in field. Interest has been growing higher-level spoken language understanding (SLU) recently, like sentiment analysis (SA). However, improving performances on SLU remains a big challenge. Basically, there two methods for tasks: (1) Two-stage method, which uses model to transfer text, then get results of downstream tasks; (2) One-stage just fine-tunes pre-trained fit tasks....