Heterogeneous remote sensing source-based change detection with optical and SAR data their combined all-time all-weather observation capability provides a reliable promising solution for wide range of applications. State-of-the-art supervised methods typically take two-stage strategy that suffers from the loss original image features introduction noise on transferred images. This paper proposes domain adaptation-based multi-source network (DA-MSCDNet) suitable to process heterogeneous...

Artificial Intelligence Machine Learning (AI/ML), in particular Deep (DL), is reorienting and transforming Earth Observation (EO). A consistent data model for delivery of training will support the FAIR principles (findable, accessible, interoperable, reusable) enable Web-based use a spatial infrastructure (SDI). Existing datasets, including open source benchmark are usually packaged into public or personal repositories lack discoverability accessibility. Moreover, there no unified method to...

Abstract The emergence of Cloud Computing technologies brings a new information infrastructure to users. Providing geoprocessing functions in platforms can bring scalable, on-demand, and cost–effective services geospatial This paper provides comparative analysis – Microsoft Windows Azure Google App Engine. compares differences the data storage, architecture model, development environment based on experience develop two platforms; emphasizes importance virtualization; recommends applications...

The existing object detection algorithm based on the deep convolution neural network needs to carry out multilevel and pooling operations entire image in order extract a semantic features of image. models can get better results for big object. However, those fail detect small objects that have low resolution are greatly influenced by noise because after repeated do not fully represent essential characteristics objects. In this paper, we achieve good accuracy extracting at different levels...

Point clouds-based 3D human pose estimation that aims to recover the locations of skeleton joints plays an important role in many AR/VR applications. The success existing methods is generally built upon large scale data annotated with joints. However, it a labor-intensive and error-prone process annotate from input depth images or point clouds, due self-occlusion between body parts as well tedious annotation on clouds. Meanwhile, easier construct datasets 2D joint annotations images. To...

Network Intrusion Detection Systems (NIDSs) play a vital role in detecting and stopping network attacks. However, the prevalent imbalance of training samples traffic interferes with NIDS detection performance. This paper proposes resampling method based on Self-Paced Ensemble Auxiliary Classifier Generative Adversarial Networks (SPE-ACGAN) to address problem sample classes. To deal class problem, SPE-ACGAN oversamples minority by ACGAN undersamples majority SPE. In addition, we merged...

Vision-based regression tasks, such as hand pose estimation, have achieved higher accuracy and faster convergence through representation learning. However, existing learning methods often encounter the following issues: high semantic level of features extracted from images is inadequate for regressing low-level information, include task-irrelevant reducing their compactness interfering with tasks. To address these challenges, we propose TI-Net, a highly versatile visual Network backbone...

Geospatial data provenance records the derivation history of a geospatial product. It is important in evaluating quality products. In Web Service environment where are often disseminated and processed widely frequently an unpredictable way, it even more identifying original sources, tracing workflows, updating or reproducing scientific results, reliability has become fundamental issue establishing spatial infrastructure (SDI). This paper investigates how to support awareness SDI. addresses...

Data management and analysis are challenging with big Earth observation (EO) data. Expanding upon the rising promises of data cubes for analysis-ready EO data, we propose a new geospatial infrastructure layered over cube to facilitate analysis. Compared previous work on cubes, proposed infrastructure, GeoCube, extends capacity multi-source vector raster GeoCube is developed in terms three major efforts: formalize dimensions process query along these dimensions, organize high-performance...

Feature and metric researchings are two vital aspects in person re-identification. Metric learning seems to have gained extra advantage over feature recent evaluations. In this paper, we explore the neglected potential of designing for We propose a novel efficient descriptor, which is motivated by traditional spatiogram covariance descriptors. The accumulates multiple spatial histograms different image regions from several color channels then extracts three descriptive sub-features. exploits...

Semantic segmentation of remote sensing (RS) images is a pivotal branch in the realm RS image processing, which plays significant role urban planning, building extraction, vegetation etc. With continuous advancement technology, spatial resolution progressively improving. This escalation gives rise to challenges like imbalanced class distributions among ground objects images, variations object scales, as well presence redundant information and noise interference. In this paper, we propose...

Agriculture is one of the most affected sectors by flood. Spaceborn remote sensing widely used for flood mapping and monitoring in recent decades. Some applications such as crop loss assessment require data with fine temporal resolution to monitor short-lived MODIS providing 1-2 days which has frequently been a large area. However, incapability penetrate through cloud hindered application optical many cases. Thus, radar especially synthetic aperture (SAR) already shows capability condition....

Medical image segmentation have drawn massive attention as it is important in biomedical analysis. Good results can assist doctors with their judgement and further improve patients' experience. Among many available pipelines medical analysis, Unet one of the most popular neural networks keeps raw features by adding concatenation between encoder decoder, which makes still widely used industrial field. In mean time, a model dominates natural language process tasks, transformer now introduced...

Remote sensing image change detection (CD) is an important task in remote analysis and essential for accurate understanding of changes the Earth’s surface. The technology deep learning (DL) becoming increasingly popular solving CD tasks images. Most existing methods based on DL tend to use ordinary convolutional blocks extract compare features, which cannot fully rich features high-resolution (HR) In addition, most lack robustness pseudochange information processing. To overcome above...

Statistics show the volume of Earth Observation (EO) data increases in exponential level during past decade. As new generation computing platform to meet big challenge, cloud significantly facilitates large-scale EO processing depending on its powerful capability. In this paper, we propose a Cloud WPS architecture integrating environment and OGC Web Services. Based architecture, implement using GeoBrain Cloud, an Apache Cloudstack based private platform, series state-of-the-art open-source...

Automatic ship detection from high-resolution optical satellite images has attracted great interest in the wide applications of maritime security and traffic control. However, most popular methods have much difficulty extracting targets without false alarms due to variable appearances ships complicated background. In this paper, we propose a approach based on visual search mechanism solve problem. First, salient regions are extracted by global contrast model fast easily. Second, geometric...

OBJECTIVE This study aimed to introduce a novel artificial intelligence (AI)–based robotic system for autonomous planning of spinal posterior decompression and verify its accuracy through cadaveric model. METHODS Seventeen vertebrae from 3 cadavers were included in the study. Three thoracic (T9–11) lumbar (L3–5) selected each cadaver. After obtaining CT data, independently planned laminectomy path based on AI algorithms before surgical procedure automatically performed during procedure. A...