A5100663938- Meteorological Phenomena and Simulations
- Precipitation Measurement and Analysis
- Air Quality and Health Impacts
- Climate variability and models
- Remote Sensing and Land Use
- GNSS positioning and interference
- Geophysics and Gravity Measurements
- Video Surveillance and Tracking Methods
- Advanced Vision and Imaging
- Air Quality Monitoring and Forecasting
- Tropical and Extratropical Cyclones Research
- Image Enhancement Techniques
- Flood Risk Assessment and Management
- Advanced Neural Network Applications
- Ocean Waves and Remote Sensing
- Atmospheric aerosols and clouds
- Advanced Measurement and Detection Methods
- Image and Signal Denoising Methods
- Human Pose and Action Recognition
- Atmospheric chemistry and aerosols
- Cryospheric studies and observations
- Advanced Image Processing Techniques
- Energy Load and Power Forecasting
- Hand Gesture Recognition Systems
- Climate Change and Health Impacts
China Meteorological Administration
2020-2025
Chengdu University of Information Technology
2016-2025
Tiangong University
2024-2025
Hangzhou Special Equipment Inspection and Research Institute
2023-2025
Lanzhou University
2015-2025
Jilin Weather Modification Office
2025
Clemson University
2024
Binghamton University
2021-2024
Huazhong University of Science and Technology
2016-2024
University of Chinese Academy of Sciences
2020-2024
The goal of precipitation nowcasting is to predict the future rainfall intensity in a local region over relatively short period time. Very few previous studies have examined this crucial and challenging weather forecasting problem from machine learning perspective. In paper, we formulate as spatiotemporal sequence which both input prediction target are sequences. By extending fully connected LSTM (FC-LSTM) convolutional structures input-to-state state-to-state transitions, propose (ConvLSTM)...
While perception tasks such as visual object recognition and text understanding play an important role in human intelligence, subsequent that involve inference, reasoning, planning require even higher level of intelligence. The past few years have seen major advances many using deep learning models. For higher-level however, probabilistic graphical models with their Bayesian nature are still more powerful flexible. To achieve integrated intelligence involves both it is naturally desirable to...
This paper reviews the NTIRE 2023 challenge on image denoising (σ = 50) with a focus proposed solutions and results. The aim is to obtain network design capable produce high-quality results best performance measured by PSNR for denoising. Independent additive white Gaussian noise (AWGN) assumed level 50. had 225 registered participants, 16 teams made valid submissions. They gauge state-of-the-art
Deep learning (DL) models have been widely used for remote sensing-based landslide mapping due to their impressive capabilities automatic information extraction. However, the large volumes of parameters and calculations compromised efficiency DL in extracting landslides from a set RS images. Lightweight convolutional neural networks (CNNs) exhibit promising feature representation abilities with fewer parameters. This study aims introduce new lightweight CNN called MS2LandsNet, designed...
To study the application of deep learning in field environmental protection, convolutional neural network VGG16 model is used to solve problem identification and classification domestic garbage. This solution first OpenCV computer vision library locate select identified objects preprocessed images into 224×224 pixel RGB accepted by network. Then after data enhancement, a based on TensorFlow framework built, using RELU activation function adding BN layer accelerate model's convergence speed,...
The Global Precipitation Measurement (GPM) mission has generated global precipitation products of improved accuracy and coverage that are promising for advanced hydrological meteorological studies. This study evaluates three Integrated Multi-satellitE Retrievals GPM (IMERG) Hourly products, including the Early-, Late-, Final-run (IMERG-HE, IMERG-HL, IMERG-HF, respectively), over Sichuan Basin China. highly complex terrain steep mountainous region offers further scrutiny on quality...
Vessel segmentation is critically essential for diagnosing a series of diseases, e.g., coronary artery disease and retinal disease. However, annotating vessel maps medical images notoriously challenging due to the tiny complex structures, leading insufficient available annotated datasets existing supervised methods domain adaptation methods. The subtle structures con-fusing background further suppress efficacy unsupervised In this paper, we propose self-supervised method via adversarial...
Recently, two types of common sensors, LiDAR and Camera, show significant performance on all tasks in 3D vision. provides accurate geometry structure, while camera captures more scene context semantic information. The fusion different sensor becomes a fundamental idea to achieve better performance. To give thorough cognition the complementary boosting about kind sensors. This paper briefly reviews enhancement systems between both sensors field depth completion, object detection, 2D\3D...
Video semantic segmentation requires to utilize the complex temporal relations between frames of video sequence. Previous works usually exploit accurate optical flow leverage relations, which suffer much from heavy computational cost. In this paper, we propose a Temporal Memory Attention Network (TMANet) adaptively integrate long-range over sequence based on self-attention mechanism without exhaustive prediction. Specially, construct memory using several past store information current frame....
The rainfall process of Chengdu region in autumn has obvious regional features. Especially, the night-time rain rate this season is very high China. Studying spatial distribution and temporal variation atmospheric precipitable water vapor (PWV) important for our understanding related processes, such as rainfall, evaporation, convective activity, among others area. Since GPS detection technology unique characteristics, all-weather, accuracy, resolution well low cost, tracking monitoring...
In this paper, we propose an automatic brain tumor segmentation algorithm based on a 22-layers deep, three-dimensional Convolutional Neural Network (CNN) for the challenging problem of gliomas segmentation. To correct bias field distortion MRI images, have added N4ITK method before intensity normalization. The use several cascaded convolution layers with small kernels allows building deeper CNN. During training, add dropout to reduce overfitting, and adapt batch normalization technique speed...
One key challenge in the point cloud segmentation is detection and split of overlapping regions between different planes. The existing methods depend on similarity dissimilarity neighbor without a global constraint, which brings 'over-' 'under-' results. Hence, this paper presents pipeline accurate plane for clouds to address shortcoming local optimization. There are two phases included proposed process. phase calculate connectivity scores planes based variations surface normals. In phase,...
Precipitation nowcasting is an important tool for weather. In recent years, progress has been achieved in some models based on deep learning precipitation nowcasting. However, these do not consider the contextual relationships between input data and output of a network their deficiency capturing information prediction objects. To overcome shortcomings, this study, we propose model that performs convolution operation Long short-term memory (LSTM) networks. Secondly, self-attention added to...
In recent years, China has suffered from frequent extreme precipitation events, and predicting their future trends become an essential part of the current research on this issue. Because inevitable uncertainties associated with individual models for climate prediction, study uses a machine learning approach to integrate fit multiple models. The results show that use several evaluation metrics provides better than traditional ensemble median method. correlation coefficients actual...
Efficiently mitigating the severe air pollution resulting from rapid progress is crucial for sustainable development of socio-ecological system. Recently, concerns about nature-based solutions have emerged in research on treatment pollution. Studies purification PM2.5 using vegetation currently concentrate individual scale tree species or urban vegetation, ignoring regional scale, which could better assist ecological governance. Therefore, taking Fenwei Plain China as study area, an...
This paper presents a novel and robust descriptor, depth-projection-map-based bag of contour fragments, which is applied to extraction hand shape structure information from depth maps. Our method projects maps onto three orthogonal planes generate the projection Then, fragment descriptors are extracted concatenated as final representation original data. A support vector machine with linear kernel used classifier. The proposed description evaluated on public datasets, well new more...
Background: Identifying Drug-Target Interactions (DTIs) is a major challenge for current drug discovery and repositioning. Compared to traditional experimental approaches, in silico methods are fast inexpensive. With the increase open-access data, numerous computational have been applied predict DTIs. Methods: In this study, we propose an end-to-end learning model of Factorization Machine Deep Neural Network (FM-DNN), which emphasizes both low-order (first or second order) high-order (higher...