A5100401385- Remote-Sensing Image Classification
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Traffic Prediction and Management Techniques
- Remote Sensing and Land Use
- Remote Sensing in Agriculture
- Advanced Image Fusion Techniques
- Web and Library Services
- Transportation Planning and Optimization
- Particle Dynamics in Fluid Flows
- Human Mobility and Location-Based Analysis
- Automated Road and Building Extraction
- Cyclone Separators and Fluid Dynamics
- Aerosol Filtration and Electrostatic Precipitation
- Smart Agriculture and AI
- Remote Sensing and LiDAR Applications
- Anomaly Detection Techniques and Applications
- Advanced Neural Network Applications
- Advanced Measurement and Detection Methods
- Infrared Target Detection Methodologies
- 3D Surveying and Cultural Heritage
- Robotics and Sensor-Based Localization
- Library Science and Administration
- Medical Image Segmentation Techniques
- Landslides and related hazards
Wuhan University
2007-2025
Beijing University of Posts and Telecommunications
2023-2024
University of Science and Technology Beijing
2024
State Key Laboratory of Information Engineering in Surveying Mapping and Remote Sensing
2020-2024
Shandong University of Technology
2023
Chengdu University of Technology
2022-2023
Northwest Normal University
2023
Tiangong University
2022
PLA Information Engineering University
2022
China University of Petroleum, East China
2019-2021
Behavior-related research areas such as motion prediction/planning, representation/imitation learning, behavior modeling/generation, and algorithm testing, require support from high-quality datasets containing interactive driving scenarios with different cultures. In this paper, we present an INTERnational, Adversarial Cooperative moTION dataset (INTERACTION dataset) in semantic maps. Five features of the are highlighted. 1) The diverse, including urban/highway/ramp merging lane changes,...
Deep learning has largely reshaped remote sensing (RS) research for aerial image understanding and made a great success. Nevertheless, most of the existing deep models are initialized with ImageNet pretrained weights since natural images inevitably present large domain gap relative to images, probably limiting fine-tuning performance on downstream scene tasks. This issue motivates us conduct an empirical study RS pretraining (RSP) images. To this end, we train different networks from scratch...
Large-scale vision foundation models have made significant progress in visual tasks on natural images, with transformers (ViTs) being the primary choice due to their good scalability and representation ability. However, large-scale remote sensing (RS) not yet been sufficiently explored. In this article, we resort plain ViTs about 100 million parameters make first attempt propose large tailored RS investigate how such perform. To handle sizes objects of arbitrary orientations a new rotated...
Hyperspectral image (HSI) classification is challenging due to spatial variability caused by complex imaging conditions. Prior methods suffer from limited representation ability, as they train specially designed networks scratch on annotated data. We propose a tri-spectral generation pipeline that transforms HSI into high-quality images, enabling the use of off-the-shelf ImageNet pretrained backbone for feature extraction. Motivated observation there are many homogeneous areas with...
In this article, we propose an end-to-end adaptive spectral-spatial multiscale network to extract contextual information for hyperspectral image (HSI) classification, which contains spectral feature extraction (FE) and spatial FE subnetworks. aspect, different from previous methods where features are obtained in a single scale, limits the accuracy improvement, two schemes based on band grouping strategy, long short-time memory (LSTM) model is used perceiving information. subnetwork,...
Foundation models have reshaped the landscape of Remote Sensing (RS) by enhancing various image interpretation tasks. Pretraining is an active research topic, encompassing supervised and self-supervised learning methods to initialize model weights effectively. However, transferring pretrained downstream tasks may encounter task discrepancy due their formulation pretraining as classification or object discrimination In this study, we explore Multi-Task (MTP) paradigm for RS foundation address...
Large-scale vision foundation models have made significant progress in visual tasks on natural images, with transformers being the primary choice due to their good scalability and representation ability. However, large-scale remote sensing (RS) not yet been sufficiently explored. In this paper, we resort plain about 100 million parameters make first attempt propose large tailored RS investigate how such perform. To handle sizes objects of arbitrary orientations a new rotated varied-size...
The success of the Segment Anything Model (SAM) demonstrates significance data-centric machine learning. However, due to difficulties and high costs associated with annotating Remote Sensing (RS) images, a large amount valuable RS data remains unlabeled, particularly at pixel level. In this study, we leverage SAM existing object detection datasets develop an efficient pipeline for generating large-scale segmentation dataset, dubbed SAMRS. SAMRS totally possesses 105,090 images 1,668,241...
In this article, we propose fully contextual networks (FullyContNets) for hyperspectral scene parsing. Different from the previous approaches that leveraging local information, proposed methods can effectively capture more generic nonlocal contexts. To end, first scale attention module (SAM) adaptively aggregate multiple features through obtaining interfeature dependencies of multiscale with self-attention mechanism, where weights are determined by measuring similarity between features. What...
This study expands on the impact of local government environmental regulation enterprise protection investment. Furthermore, it analyzes influence promotion pressure officials has scale The results show that investment companies in China is generally insufficient. attitude toward passive under policy regulation. supervision also still at a low level. Both these observations are far from intentions government. There U-shaped relationship between official and Only when exceeds certain limit...
We present Hunyuan3D 2.0, an advanced large-scale 3D synthesis system for generating high-resolution textured assets. This includes two foundation components: a shape generation model -- Hunyuan3D-DiT, and texture Hunyuan3D-Paint. The generative model, built on scalable flow-based diffusion transformer, aims to create geometry that properly aligns with given condition image, laying solid downstream applications. benefiting from strong geometric priors, produces vibrant maps either generated...
Jailbreak attacks against large language models (LLMs) aim to induce harmful behaviors in LLMs through carefully crafted adversarial prompts. To mitigate attacks, one way is perform training (AT)-based alignment, i.e., on some of the most prompts help them learn how behave safely under attacks. During AT, length plays a critical role robustness aligned LLMs. This paper focuses suffix jailbreak and unveils that defend attack with an $\Theta(M)$, it enough align suffixes $\Theta(\sqrt{M})$....
Convolutional neural networks (CNNs) have been widely applied to hyperspectral image classification (HSIC). However, traditional convolutions can not effectively extract features for objects with irregular distributions. Recent methods attempt address this issue by performing graph on spatial topologies, but fixed structures and local perceptions limit their performances. To tackle these problems, in article, different from previous approaches, we perform the superpixel generation...
Recent years have witnessed the superiority of deep learning-based algorithms in field HSI classification. However, a prerequisite for favorable performance these methods is large number refined pixel-level annotations. Due to atmospheric changes, sensor differences, and complex land cover distribution, labeling high-dimensional hyperspectral image (HSI) extremely difficult, time-consuming, laborious. To overcome above hurdle, an Image-To-pixEl Representation (ITER) approach proposed this...
Numerous aquaculture ponds are intensively distributed around inland natural lakes and mixed with cropland, especially in areas high population density Asia. Information about the distribution of is essential for monitoring impact human activities on lakes. Accurate efficient mapping using high-spatial-resolution remote-sensing images a challenging task because mingled other land cover types. Considering that have intertwining regular embankments these salient features prominent at different...
Landcover classification is an important application in remote sensing, but it always a challenge to distinguish different features with similar characteristics or large-scale differences. Some deep learning networks, such as UperNet, PSPNet, and DANet, use pyramid pooling attention mechanisms improve their abilities multi-scale extraction. However, due the neglect of low-level contained underlying network information differences between feature maps, difficult identify small-scale objects....