Road extraction from remote sensing images in very high resolution is important for autonomous driving and road planning. Compared with large-scale objects, roads are smaller, winding, likely to be covered by buildings' shadows, causing deep convolutional neural networks (DCNNs) difficult identify roads. The paper proposes a semantics-geometry framework (SGNet) two-branch backbone, i.e., semantics-dominant branch geometry-dominant branch. inputs predict dense semantic features, the takes...

Canopy architecture determines the light distribution and interception in canopy. Reasonable shaping pruning can optimize tree structure; maximize utilization of land, space energy; lay foundation for achieving early fruiting, high yield, health longevity. Due to complexity loquat canopy multi-year period growth, variables needed experiments type training are hardly accessible through field measurements. In this paper, we concentrated on exploring relationship between branching angle using a...

Artificial Intelligence (AI) Machine Learning (ML) technologies, particularly Deep (DL), have demonstrated significant potential in the interpretation of Remote Sensing (RS) imagery, covering tasks such as scene classification, object detection, land-cover/land-use change and multi-view stereo reconstruction. Large-scale training samples are essential for ML/DL models to achieve optimal performance. However, current organization is ad-hoc vendor-specific, lacking an integrated approach that...

The early warning of fires is pivotal for preventing substantial economic losses and ecological damage. However, the enhancement fire model generalization performance still faces challenges due to limitations in existing datasets, such as image quantity heterogeneity. Taking advantage rapid advancements remote sensing technology artificial intelligence, we introduce Flame And Smoke Detection Dataset (FASDD), a pioneering collection comprising over 120,000 heterogeneous images covering...

Abstract. A tailing dam accident can cause serious ecological disaster and property loss. Simulation of a in advance is useful for understanding the flow characteristics assessing possible extension impact area. In this paper, three-dimensional (3-D) computational fluid dynamics (CFD) approach was proposed reasonably quickly predicting routing area mud from failure across 3-D terrain. The Navier–Stokes equations Bingham–Papanastasiou rheology model were employed as governing constitutive...

Deep learning-based fire detection models are usually trained offline on static datasets. For continuously increasing heterogeneous sensor data, incremental learning is a resolution to enable updates of models. However, it still encounters the challenge stability-plasticity dilemma cross-domain data. In this paper, we propose Dynamic Equilibrium Network (DENet) achieve images captured by spaceborne, airborne, and terrestrial sensors. It can learn more flame smoke features from data in...

Airborne platforms have been improved in the past decade to provide geographic information systems (GISs) with large-scale 3D geographical information. Objectification of such organized meshes is a significant challenge for GISs. The ground filtering key step meeting this challenge, however, its accuracy highly affected by negative blunders and unbalanced vertex density. This paper proposes novel method differentiating geometric primitives from realistic based on cloth simulation filter....

Abstract. With the advancement of computer vision, artificial intelligence, and remote sensing technologies, deep learning algorithms are increasingly used in terrestrial, airborne, spaceborne-based fire detection systems. The performance generalization these data-driven algorithms, however, restricted by limited number source datasets. A large-scale benchmark dataset covering complex varied scenarios is urgently needed. This work constructs a 100,000-level Flame Smoke Detection Dataset...

It is of great significance for disaster prevention and mitigation to carry out simulations dam failure accidents in advance, but at present, there are few professional systems tailings dams. In this paper, we focused on the construction a virtual geographic environment (VGE) system that provides an effective tool visualizing dam-break process pond. The numerical model based computational fluid dynamics (CFD) was integrated into VGE system. infrastructure supported by 3-D information (GIS)...

Recently, deep learning has been widely used in the segmentation tasks of remote sensing images. However, existing method most focus on local contextual information and limited field perception, which makes it difficult to capture long-range feature objects at large scales form very-high-resolution (VHR) In this paper, we present a novel Local–global Framework consisting dual-source fusion network local–global transformer modules, efficiently utilize features extracted from multiple sources...

Taxi demand forecasting plays an important role in ride-hailing services. Accurate taxi can assist companies pre-allocating taxis, improving vehicle utilization, reducing waiting time, and alleviating traffic congestion. It is a challenging task due to the highly non-linear complicated spatial-temporal patterns of data. Most existing methods lack ability capture dynamic dependencies among regions. They either fail consider limitations Graph Neural Networks or do not efficiently long-term...

Transitions of spatial data infrastructures (SDIs) support applications from 2D landscapes to 3D scenes. The existing methods for describing, managing, and providing services often lack coordination efficiency. Moreover, the added complexity structures necessitates novel approaches component-level management streaming capabilities. In response, we developed a generic conceptual model suitable diverse in SDIs discussed design rationales key considerations underlying model. We formalized...

Most deep-learning methods that achieve high segmentation accuracy require deep network architectures are too heavy and complex to run on embedded devices with limited storage memory space. To address this issue, letter proposes an efficient generative adversarial transformer (GATrans) for achieving high-precision semantic while maintaining extremely size. The framework utilizes a global (GTNet) as the generator, efficiently extracting multilevel features through residual connections. GTNet...

Abstract. Deep learning methods driven by in situ video and remote sensing images have been used fire detection. The performance generalization of detection models, however, are restricted the limited number modality training datasets. A large-scale benchmark dataset covering complex varied scenarios is urgently needed. This work constructs a 100,000-level Flame Smoke Detection Dataset (FASDD) based on multi-source heterogeneous flame smoke images. To best our knowledge, FASDD currently most...

Snow and ice albedo is a critical geographical indicator that reflects climate change on Earth. Quantifying the in Greenland sheet, which extensively covered with snow ice, key to studying changes energy budget Northern Hemisphere. Earth observation satellites have been regularly providing surface products. However, optical satellite-derived products many voids due persistent cloud cover over sheet. Consequently, seamless reconstruction of spatial temporal scales essential. Surface albedo,...

Abstract. A tailings dam accident can cause serious ecological disaster and property loss. Simulation of a in advance is useful for understanding the flow characteristics assessing possible extension impact area. In this paper, three-dimensional (3-D) computational fluid dynamics (CFD) approach was proposed reasonably quickly predicting routing area mud from failure across 3-D terrain. The Navier–Stokes equations Bingham-Papanastasiou rheology model were employed as governing constitutive...

In recent years, the availability of online geospatial analysis tools has increased, allowing users to access and manipulate large datasets perform complex analyses through Web-based interfaces. It is still a challenge on how manage resources including data functions in distributed infrastructure. To address this issue, study proposes model hub approach for analysis. Two core components, named Data Center Model Center, are involved. The serves as integrating, storing, sharing metadata....

The spatiotemporally continuous data of normalized difference snow index (NDSI) are key to understanding the mechanisms occurrence and development as well patterns distribution changes. However, presence clouds, particularly prevalent in polar regions such Greenland Ice Sheet (GrIS), introduces a significant number missing pixels MODIS NDSI daily data. To address this issue, study proposes utilization spatiotemporal extreme gradient boosting (STXGBoost) model generate comprehensive dataset....