A5100324309- Advanced Power Amplifier Design
- Radio Frequency Integrated Circuit Design
- PAPR reduction in OFDM
- Wireless Communication Networks Research
- Human Pose and Action Recognition
- Generative Adversarial Networks and Image Synthesis
- Advanced Image Processing Techniques
- Optical Systems and Laser Technology
- Advanced Wireless Communication Techniques
- Advanced Image and Video Retrieval Techniques
- Anomaly Detection Techniques and Applications
- Direction-of-Arrival Estimation Techniques
- Full-Duplex Wireless Communications
- Seismic Imaging and Inversion Techniques
- Image Processing Techniques and Applications
- Advanced Measurement and Detection Methods
- Advanced Adaptive Filtering Techniques
- Advanced Vision and Imaging
- Metallurgy and Material Forming
- Wireless Power Transfer Systems
- Speech and Audio Processing
- Higher Education and Teaching Methods
- Image Retrieval and Classification Techniques
- Image and Signal Denoising Methods
- Electrical Contact Performance and Analysis
Xidian University
2009-2025
Clemson University
2024
Northeastern University
2023-2024
Technical University of Darmstadt
2024
Institute of Electrical Engineering
2024
Baidu (China)
2018-2023
China Geological Survey
2023
Washington University in St. Louis
2023
Portland State University
2010-2022
National Space Science Center
2022
The problem of video classification is inherently sequential and multimodal, deep neural models hence need to capture aggregate the most pertinent signals for a given input video. We propose Keyless Attention as an elegant efficient means more effectively account nature data. Moreover, comparing variety multimodal fusion methods, we find that Multimodal Fusion successful at discerning interactions between modalities. experiment on four highly heterogeneous datasets, UCF101, ActivityNet,...
Artistic style transfer aims at migrating the from an example image to a content image. Currently, optimization-based methods have achieved great stylization quality, but expensive time cost restricts their practical applications. Meanwhile, feed-forward still fail synthesize complex style, especially when holistic global and local patterns exist. Inspired by common painting process of drawing draft revising details, we introduce novel method named Laplacian Pyramid Network (LapStyle)....
Human pose transfer has received great attention due to its wide applications, yet is still a challenging task that not well solved. Recent works have achieved success the person image from source target pose. However, most of them cannot capture semantic appearance, resulting in inconsistent and less realistic textures on reconstructed results. To address this issue, we propose new two-stage framework handle appearance translation. In first stage, predict parsing maps eliminate difficulties...
This paper reviews the NTIRE 2020 challenge on perceptual extreme super-resolution with focus proposed solutions and results.The task was to super-resolve an input image a magnification factor ×16 based set of prior examples low corresponding high resolution images.The goal is obtain network design capable produce results best quality similar ground truth.The track had 280 registered participants, 19 teams submitted final results.They gauge state-of-the-art in single superresolution....
In this brief an efficient folded architecture (EFA) for lifting-based discrete wavelet transform (DWT) is presented. The proposed EFA based on a novel form of the lifting scheme that given in brief. Due to form, conventional serial operations data flow can be optimized into parallel ones by employing and pipeline techniques. corresponding (OA) has short critical path latency repeatable. Further, utilizing repeatability, derived from OA fold technique. For EFA, hardware utilization achieves...
This paper describes our solution for the video recognition task of ActivityNet Kinetics challenge that ranked 1st place. Most existing state-of-the-art approaches are in favor an end-to-end pipeline. One exception is framework DevNet. The merit DevNet they first use data to learn a network (i.e. fine-tuning or training from scratch). Instead directly using classification scores (e.g. softmax scores), extract features learned and then fed them into off-the-shelf machine learning models...
This paper describes our solution for the video recognition task of Google Cloud and YouTube-8M Video Understanding Challenge that ranked 3rd place. Because challenge provides pre-extracted visual audio features instead raw videos, we mainly investigate various temporal modeling approaches to aggregate frame-level multi-label recognition. Our system contains three major components: two-stream sequence model, fast-forward model residual neural networks. Experiment results on challenging...
Recent generative adversarial network based methods have shown promising results for the charming but challenging task of synthesizing images from text descriptions. These approaches can generate with general shape and color often produce distorted global structures unnatural local semantic details. It is due to ineffectiveness convolutional neural networks in capturing high-level information pixel-level image synthesis. In this paper, we propose a Dual Attentional Generative Adversarial...
Lithium–sulfur batteries (LSBs) have been considered as potential next-generation energy storage systems due to their high specific of 2600 Wh kg–1 and 2800 L–1. Nevertheless, the practical application LSBs still faces several hazards, including shuttle effect soluble lithium polysulfides, low electrical conductivities solid sulfur sulfides, large volume expansion during charge/discharge cycles. To address this critical challenge, we innovatively proposed facile synthesis nanostructured VN...
Text-driven image manipulation remains challenging in training or inference flexibility. Conditional generative models depend heavily on expensive annotated data. Meanwhile, recent frameworks, which leverage pre-trained vision-language models, are limited by either per text-prompt optimization inference-time hyper-parameters tuning. In this work, we propose a novel framework named DeltaEdit to address these problems. Our key idea is investigate and identify space, namely delta text space...
Nitrites (NO
Hyperspectral video acquisition requires a precise balance between spectral and temporal resolution, often achieved through compressive sampling using two-dimensional detectors reconstruction algorithms. However, the reliance on spatial light modulators for coding reduces optical efficiency, while complex recovery algorithms hinder real-time reconstruction. To address these challenges, we propose digital-micromirror-device-based complementary dual-channel hyperspectral (DMD-CDH) imaging...
In remote sensing image scene classification (RSISC) tasks, downsampling is crucial for reducing computational complexity and cache demands, enhancing the model’s generalization capability of deep neural networks. Traditional methods, such as regular fixed lattice approaches (pooling in CNN token merging transformers), often flatten distinguishing texture features, impacting performance. To address this, we propose an adaptive transformer (ATMformer) that preserves essential local features...
Generative Artificial Intelligence (GAI) is an emerging and disruptive technology that has attracted considerable interest from researchers educators across various disciplines. We discuss the relevance concerns of ChatGPT other GAI tools in environmental psychology research. propose three use categories for tools: integrated contextualized understanding, practical flexible implementation, two-way external communication. These are exemplified by topics such as health benefits green space,...
Existing action localization approaches adopt shallow temporal convolutional networks (i.e., TCN) on 1D feature map extracted from video frames. In this paper, we empirically find that stacking more conventional convolution layers actually deteriorates classification performance, possibly ascribing to all channels of map, which generally are highly abstract and can be regarded as latent concepts, excessively recombined in convolution. To address issue, introduce a novel concept-wise network...
Aiming at the poor adaptability of expert system in air combat,a maneuvering decision algorithm based on receding horizon control( RHC) method was proposed to improve combat decision-making system. Firstly,the optimal control problem systematically analyzed The state equation,the index function and constraints model were established. On this basis,according principle RHC method,the whole process divided into some sequential ones with finite time horizon. In each horizon,the decisionmaking...
Although orthogonal frequency division multiplexing (OFDM) has been standardized for 5G, filter bank multi-carrier (FBMC) and filtered (F-OFDM) remain competitive as candidates future generations of wireless technologies beyond due to their reduced spectrum leakage thus enhanced efficiency. In this article, we developed a unified expression OFDM, FBMC, F-OFDM, which provides comparative insights into those techniques. A representative sideband quantification is included at the end article.