A5100649796- Automated Road and Building Extraction
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Remote Sensing and LiDAR Applications
- Remote-Sensing Image Classification
- Video Surveillance and Tracking Methods
- Medical Image Segmentation Techniques
- Image and Signal Denoising Methods
- Remote Sensing and Land Use
- Advanced Neural Network Applications
- 3D Surveying and Cultural Heritage
- Urban Heat Island Mitigation
- Biomedical Text Mining and Ontologies
- Semantic Web and Ontologies
- Advanced Measurement and Detection Methods
- Brain Tumor Detection and Classification
- Advanced Image and Video Retrieval Techniques
- Land Use and Ecosystem Services
- Advanced Image Fusion Techniques
- Satellite Image Processing and Photogrammetry
- AI in cancer detection
- Human Pose and Action Recognition
- Data Visualization and Analytics
- Industrial Vision Systems and Defect Detection
- Medical Imaging and Analysis
- Stroke Rehabilitation and Recovery
Hangzhou Dianzi University
2010-2024
Technical University of Munich
2016-2024
Harbin Engineering University
2004-2022
Sichuan University of Arts and Science
2022
Southwest Minzu University
2022
Xi’an Jiaotong-Liverpool University
2022
Traffic Management Research Institute
2022
Deutsches Zentrum für Luft- und Raumfahrt e. V. (DLR)
2017-2021
Mudanjiang Normal University
2021
Harbin Institute of Technology
2020
Gaining access to labeled reference data is one of the great challenges in supervised machine-learning endeavors. This especially true for an automated analysis remote sensing images on a global scale, which enables us address challenges, such as urbanization and climate change, using state-of-the-art techniques. To meet these pressing needs, urban research, we provide open valuable benchmark set, So2Sat LCZ42, consists local-climate-zone (LCZ) labels approximately half million...
Background: Action observation (AO) has the potential to improve motor imagery (MI) practice in stroke patients. However, currently only a few results are available on how use AO effectively.Objective: The aim of this study is investigate whether MI can be improved more effectively by synchronous than asynchronous AO.Methods: Ten patients with upper limb dysfunction following were selected as participants. They divided into two groups perform combined daily conventional rehabilitation for...
Rule-based modeling (RBM) is a powerful and increasingly popular approach to cell signaling networks. However, novel visual tools are needed in order make RBM accessible broad range of users, specification models less error prone, improve workflows. We introduce RuleBender, visualization system for the integrated visualization, simulation rule-based intracellular biochemistry. present user requirements, paradigms, algorithms design decisions behind with emphasis on global/local model...
Building height retrieval from synthetic aperture radar (SAR) imagery is of great importance for urban applications, yet highly challenging due to the complexity SAR data. This paper addresses issue building in large-scale areas a single TerraSAR-X spotlight or stripmap image. Based on viewing geometry, we propose that this problem be formulated as bounding box regression and therefore allows integrating data multiple sources generating ground truth larger scale. We introduce footprints...
Access to labeled reference data is one of the grand challenges in supervised machine learning endeavors. This especially true for an automated analysis remote sensing images on a global scale, which enables us address such as urbanization and climate change using state-of-the-art techniques. To meet these pressing needs, urban research, we provide open access valuable benchmark dataset named "So2Sat LCZ42," consists local zone (LCZ) labels about half million Sentinel-1 Sentinel-2 image...
Object retrieval and reconstruction from very-high-resolution (VHR) synthetic aperture radar (SAR) images are of great importance for urban SAR applications, yet highly challenging due to the complexity data. This article addresses issue individual building segmentation a single VHR image in large-scale areas. To achieve this, we introduce footprints geographic information system (GIS) data as complementary propose novel conditional GIS-aware network (CG-Net). The proposed model learns...
Three-dimensional (3D) building structures are vital to understanding urban dynamics. Monocular remote sensing imagery is a cost-effective data source for large-scale height retrieval when compared LiDAR and multi-view imagery. Existing methods learn footprints maps per pixel via either multi-task network or two separate networks, however, failing consider the information of neighboring pixels that belong identical building. Therefore, we propose learning novel representation 3D buildings,...
Quick and automated earthquake-damaged building detection from post-event satellite imagery is crucial, yet it challenging due to the scarcity of training data required for developing robust algorithms. This letter presents first dataset dedicated detecting buildings very high resolution (VHR) Synthetic Aperture Radar (SAR) optical imagery. Utilizing open annotations acquired after 2023 Turkey–Syria earthquakes, we deliver a co-registered footprints image patches both SAR data, encompassing...
Building segmentation is of great importance in the task remote sensing imagery interpretation. However, existing semantic and instance methods often lead to masks with blurred boundaries. In this paper, we propose a novel network for building high-resolution images. More specifically, consider segmenting an individual as detecting several keypoints. The detected keypoints are subsequently reformulated closed polygon, which boundary building. By doing so, sharp could be preserved....
Existing techniques of 3-D reconstruction buildings from SAR images are mostly based on multibaseline interferometry, such as PSI and tomography (TomoSAR). However, these require tens for a reliable reconstruction, which limits the application in various scenarios, emergency response. Therefore, alternatives that use single image building footprints GIS data show their great potential reconstruction. The combination requires precise registration, is challenging due to unknown terrain height,...
Higher standards have been proposed for detection systems since camouflaged objects are not distinct enough, making it possible to ignore the difference between their background and foreground. In this paper, we present a new framework Camouflaged Object Detection (COD) named FSANet, which consists mainly of three operations: spatial detail mining (SDM), cross-scale feature combination (CFC), hierarchical aggregation decoder (HFAD). The simulates three-stage process human visual mechanism...
The precise segmentation of organs at risk (OARs) is importance for improving therapeutic outcomes and reducing injuries patients undergoing radiotherapy. In this study, we developed a new approach accurate computed tomography (CT) image the eyes surrounding organs, which first locating then (FLTS).The FLTS was composed two steps: (a) classification CT images using convolutional neural networks (CNN), (b) modified U-shape networks. order to obtain optimal performance, enhanced our training...
Accurate annotation of the medical image is crucial step for artificial intelligence (AI) clinical application. However, annotating will incur a lot efforts and expense due to its high complexity needing experienced doctors. In order reduce cost annotation, several active learning methods have been proposed previously. focus these number candidates with little regard doctor's workload, which not enough since even small amount data take time effectively workload doctors, we developed new...
Height reconstruction of large-scale buildings from single very high resolution (VHR) SAR image is great interest especially in applications with temporal restrictions. The problem highly challenging due to the inherent complexity images, e.g., side-looking geometry and different microwave scattering contributions. In this work, we present a framework estimate building heights VHR image. individual are defined by GIS data, deep neural network used segment wall area layover length then...
Automatic detection and reconstruction of man-made objects from single very high resolution (VHR) SAR image is great interest especially when it comes to applications having stringent temporal restrictions, e.g., emergency responses. In particular, estimation building height key issue in such post scenarios aiding information retrieval. However, due inherent complexity images originating side-looking geometry different microwave scattering contributions, the problem highly challenging. this...
3D building reconstruction from monocular remote sensing imagery is a promising and economical way to generate city models at large scale, yet the task rarely touched. The paper tackles problem via an end-to-end network. goal achieved by modified network, named Mask-Height R-CNN, based on Mask with additional height prediction head in Region Proposal Network (RPN). Unlike most deep learning methods, estimation done instance level instead of pixel level, which does not require assembly maps...
Accurate segmentation of MR brain tissue is a crucial step for diagnosis,surgical planning, and treatment abnormalities. However,it time-consuming task to be performed by medical experts. So, automatic reliable methods are required. How choose appropriate training dataset from limited labeled rather than the whole also has great significance in saving time. In addition, data too rare expensive obtain extensively, so choosing unlabeled instead all datasets annotate, which can attain at least...
We present a mathematical model of multimedia data streams and framework for functional dependency analysis. The dual objectives are to effectively design schema efficiently process continuous queries on sensor-based streams. To further improve query processing, we introduce the concept ontological filtering. A software tool add dependencies ontology is developed. Based upon analysis filtering, processing algorithms, illustrative examples experimental results querying presented demonstrate...
Noticeable progress has been witnessed in general object detection, semantic segmentation and instance segmentation, while parsing a group of people is still challenging task for human-centric visual understanding due to severe occlusion various poses. In this paper, we present new large-scale dataset named "GPS (Group People Segmentation)" boost academical study technology development. GPS contains 14000 elaborately annotated images with 20 fine-grained category labels related human,...