A5026770433- Robotics and Sensor-Based Localization
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Advanced Image and Video Retrieval Techniques
- Autonomous Vehicle Technology and Safety
- Remote Sensing and LiDAR Applications
- 3D Shape Modeling and Analysis
- 3D Surveying and Cultural Heritage
- Human Pose and Action Recognition
- Multimodal Machine Learning Applications
- Computer Graphics and Visualization Techniques
- Domain Adaptation and Few-Shot Learning
- Visual Attention and Saliency Detection
- Image Retrieval and Classification Techniques
- Automated Road and Building Extraction
- Numerical methods in engineering
- Simulation and Modeling Applications
- Composite Structure Analysis and Optimization
- Generative Adversarial Networks and Image Synthesis
- Traffic Prediction and Management Techniques
- Image and Object Detection Techniques
- Sparse and Compressive Sensing Techniques
- Ultrasonics and Acoustic Wave Propagation
- Medical Image Segmentation Techniques
Xi'an Jiaotong University
2016-2025
Southwest Petroleum University
2024
Northwestern Polytechnical University
2023
Shantou University
2007-2022
Shantou University Medical College
2007-2022
Cancer Hospital of Shantou University Medical College
2022
Huazhong Agricultural University
2020
Shanghai Tunnel Engineering Rail Transit Design & Research Institute
2018
Tongji University
2002-2015
Hebei Medical University
2007
3D lane detection which plays a crucial role in vehicle routing, has recently been rapidly developing topic autonomous driving. Previous works struggle with practicality due to their complicated spatial transformations and inflexible representations of lanes. Faced the issues, our work proposes an efficient robust monocular called BEV-LaneDet three main contributions. First, we introduce Virtual Camera that unifies in/extrinsic parameters cameras mounted on different vehicles guarantee...
Abstract Developing ambient‐air fabrication strategy is desirable for lowering down the cost of perovskite light emitting diodes (LEDs) and further promoting their broad applications. However, ambient humidity usually leads to undesirable interface film quality, causing severe nonradiative losses unsatisfactory device performance. In this work, an effective solve problem remarkably enhance performance LEDs reported. The study reveals that humidity‐induced aggregation self‐assembled monolayer...
Introduction Many plants can modify their leaf profile rapidly in response to environmental stress. Image-based data are increasingly used retrieve reliable information on plant water status a non-contact manner that has the potential be scaled high-throughput and repeated through time. This paper examined variation of angle as measured by both 3D images goniometer progressively drought stressed grapevine. Grapevines, grown pots, were subjected 21-day period stress receiving 100% (CTRL), 60%...
In this paper, we present a novel framework to allow users tour simulated traffic scenes from the first-person view. Constructing 3-D road image sequences is in general difficult, due intrinsic complexity of dynamic scenes, which are composed drastically moving background, not mention numerous other surrounding vehicles. With definitions scene models, first introduce construction process simple scenes. After detection boundaries by semantic fast two-cycle (FTC) level set method, generate...
Scene Graph Generation (SGG) plays an important role in enhancing visual image comprehension. However, existing approaches often struggle to represent implicit relationship features, resulting a limited ability distinguish predicates. Meanwhile, they are vulnerable skewed instance distributions, which impairs effective training for fine-grained To address these problems, we propose novel feature refinement and data redistribution framework (RAR). Specifically, multi-domain fusion (MDF)...
Robust time and frequency synchronization is an essential requirement for initial access in LEO satellite communication systems. However, large doppler shift system severely influences the accuracy of timing estimation with Zadoff-Chu sequence. In this paper, improved detection method proposed system, which can achieve accurate performance presence shift.
Human can judge driver's driving ability by observing the vehicle motion in different traffic scenes. Identically, behavior be main basis for evaluating performance of an unmanned both field test and simulation test. Although avoids disadvantages test, existing systems lack scene data with perception granularity visual sensors. In order realizing vehicle-in-loop simulation, technique behaviors must able to exhibit actual vehicles. this paper, we propose automatic approach simulating...
Leveraging the consensus and complementary principle to find a common representation for different views is an essential problem of multi-view clustering. To address problem, many Low-Rank Representation (LRR) based methods have been proposed. However, existing LRR two limitations: 1) they adopt graph regularization that only considers simple pairwise similarities among data points, 2) do not generally characterize importance each view. In this letter, we correspondingly utilize hypergraph...
In this traffic-scene-modeling study, we propose an image-captioning network which incorporates element attention into encoder-decoder mechanism to generate more reasonable scene captions. A visual-relationship-detecting is also developed detect the relative positions of object pairs. Firstly, traffic elements are detected and segmented according their clustered locations. Then, applied corresponding description each element. The utilized position relations all pairs in subregion. static...
As a critical transformational process in the attributes of epithelial cells, epithelial-to-mesenchymal transition (EMT) is involved tumor invasion, metastasis, and resistance to treatment, which contributes ultimate death some patients with breast cancer. Glycogen synthase kinase-3-beta (GSK3β) thought be an EMT suppressor that down-regulates protein, snail, zinc finger transcription inhibitor, regulates E-cadherin expression Wnt signaling pathway. Our previous studies have shown Notch3...
Since Jacquin proposed original fractal image compression technique in 1990, coding method has been developed into various schemes. Traditionally, uses mean square error (MSE) to evaluate similarity of blocks, but the evaluated by MSE usually differs from human visual system (HVS). Compared with MSE, structural (SSIM) is an measure index which more appropriate for HVS. This paper proposes a new scheme between blocks and compute these blocks' coefficients. The experiment results show that...
Although the topic of image search has been heavily studied in last two decades, many works have focused on either instance-level retrieval or semantic-level retrieval. In this work, we develop a novel visually similar spatial-semantic method, namely spatial-content search, to images that not only share same spatial-semantics but also enjoy visual consistency as query complex scenes. We achieve goal by capturing concepts well representation each concept contained an image. Specifically,...
For the task of unsupervised image translation, transforming style while preserving its original structure remains challenging. In this paper, we propose an translation method with structural enhancement in frequency domain named SEIT. Specifically, a dynamic adaptive (FDA) module is designed for transformation that can well transfer maintaining overall by decoupling content and domain. Moreover, wavelet-based (WSE) proposed to improve intermediate results matching high-frequency...
Recently, the low‐rank and sparse (LRS) matrix decomposition has been introduced as an effective mean to solve multi‐view registration. It views each available relative motion a block element reconstruct one matrix, which then is used approximate where global motions can be recovered for However, this approach sensitive sparsity of reconstructed it treats all elements equally in spite their varied reliabilities. Therefore, study proposes registration by weighted LRS decomposition. On basis...
A new framework to construct 3D traffic scenes for unmanned vehicle simulation evaluation is proposed in this paper. The "floor-wall" geometry applied construction process which consists of two stages. first stage road planes specification implemented by support vector machine (SVM) and Markov random field (MRF) using superpixels. In the second stage, control nodes boundaries are specified background scenes. Foreground elements supplemented, assumed stand perpendicularly planes. New...