A5101620919- Remote-Sensing Image Classification
- Advanced Image Fusion Techniques
- Infrastructure Maintenance and Monitoring
- Remote Sensing and LiDAR Applications
- Automated Road and Building Extraction
- Landslides and related hazards
- Dynamics and Control of Mechanical Systems
- Spectroscopy and Chemometric Analyses
- Video Surveillance and Tracking Methods
- Geochemistry and Geologic Mapping
- Border Security and International Relations
- Advanced Image and Video Retrieval Techniques
- Advanced Measurement and Detection Methods
- Indoor and Outdoor Localization Technologies
- Image Enhancement Techniques
- Face and Expression Recognition
- Target Tracking and Data Fusion in Sensor Networks
- Spinal Fractures and Fixation Techniques
- Fire Detection and Safety Systems
- Satellite Communication Systems
- Opportunistic and Delay-Tolerant Networks
- Remote Sensing in Agriculture
- Technology and Security Systems
- Geographic Information Systems Studies
- Smart Grid and Power Systems
Shandong Institute of Business and Technology
2023-2024
Beihang University
2024
Dongguan People’s Hospital
2000
Landslide, a kind of destructive natural disaster, often occurs in the mountainous areas China. Landslide information instant collection plays an important role taking appropriate remedial measures and personnel evacuation. In recent years, use Convolutional Neural Network (CNN) for landslide regional detection achieved good performance, however, most CNN-based methods had no regard internal connection cover materials disaster occurrence area. Moreover, revealed by deformation features was...
Graph Neural Networks (GNNs), as an effective learning framework for graph structure data representation, have been applied to hyperspectral images (HSIs) classification tasks. Among the variants of GNNs, Attention (GATs) achieved state-of-the-art node prediction performance by assign dense attention coefficients all neighbors feature aggregation. However, due complexity land distribution and high dimension HSIs data, it is difficult identify different coverage categories employing GATs...
Map construction task plays a vital role in providing precise and comprehensive static environmental information essential for autonomous driving systems. Primary sensors include cameras LiDAR, with configurations varying between camera-only, LiDAR-only, or camera-LiDAR fusion, based on cost-performance considerations. While fusion-based methods typically perform best, existing approaches often neglect modality interaction rely simple fusion strategies, which suffer from the problems of...
Abstract Collaborative robots are becoming intelligent assistants of human in industrial settings and daily lives. Dynamic model identification is an active topic for collaborative because it can provide effective ways to achieve precise control, fast collision detection smooth lead-through programming. In this research, improved iterative approach with a comprehensive friction dynamic proposed when the joint velocity, temperature load torque effects considered. Experiments conducted on AUBO...
The road segmentation task has become increasingly important in fields such as urban planning, traffic management, and environmental monitoring. However, most existing deep learning-based methods suffer from issues poor temporal effectiveness connectivity, making it a significant challenge to achieve high-precision high-efficiency segmentation. We propose model based on detail-enhanced lightweight transformer. Through the connectivity enhancement module, issue of spatial information loss is...
The excellent capabilities of Transformers and Graph Neural Networks (GNNs) in modelling long-range dependencies handling irregular data have led to their widespread application hyperspectral image (HSI) classification tasks. However, the Transformer combining both advantages is rarely used this field has some limitations. Current consider interactions between all nodes within graph, adding complexity introducing unnecessary information from noisy nodes. Moreover, rich spectral HSIs often...
Target localization is one of the typical applications wireless sensor networks. However, there are few algorithms addressing problems caused by both noisy channels and limited sensing range energy nodes. In this paper, a new source algorithm based on maximum likelihood (ML) estimation incorporating channel impact proposed. The Cramer-Rao bound (CRB) ML acoustic location estimate has been derived to analyze accuracy estimations. Extensive simulations have conducted. results show that...
As an important feature element in remote sensing images, road network extraction images has become a research hotspot for scholars at home and abroad. With the development of deep learning, based on convolutional neural achieved certain results. However, due to complexity environment, it is still difficult quickly accurately separate content from surrounding land cover high-resolution images. In this paper, we make improvements LinkNet address above problems: firstly, coding layer part,...
To strengthen the communication data carrying capacity of electric power local area networks and achieve a relatively stable connection form in entire network, an intelligent perception system for security situations is designed. Based on B/S framework system, components within network are divided into layered forms, external interfaces utilized to establish connections between modules. On this basis, server-side thread established obtain accurate information parsing results through...
Faced with the problems of high energy consumption and incomplete data extraction in multi-feature fusion, a fusion method based on depth autoencoder is proposed. A constructed to extract transmission lines. To solve problem gradient disappearance, extracted trained by layer-by-layer training method. The forward propagation used calculate confidence judge stability sensor cluster. similarity between different features calculated using Babbitt coefficient, weights are obtained weighted...