scholarly journals Object Detection and Image Segmentation with Deep Learning on Earth Observation Data: A Review—Part II: Applications

2020 ◽  
Vol 12 (18) ◽  
pp. 3053 ◽  
Author(s):  
Thorsten Hoeser ◽  
Felix Bachofer ◽  
Claudia Kuenzer
Keyword(s):  
Image Segmentation ◽  
Deep Learning ◽  
Object Detection ◽  
Land Surface ◽  
Large Scale ◽  
Earth Observation ◽  
Observation Data ◽  
Fine Grained ◽  
The Impact ◽  
Object Dynamics

In Earth observation (EO), large-scale land-surface dynamics are traditionally analyzed by investigating aggregated classes. The increase in data with a very high spatial resolution enables investigations on a fine-grained feature level which can help us to better understand the dynamics of land surfaces by taking object dynamics into account. To extract fine-grained features and objects, the most popular deep-learning model for image analysis is commonly used: the convolutional neural network (CNN). In this review, we provide a comprehensive overview of the impact of deep learning on EO applications by reviewing 429 studies on image segmentation and object detection with CNNs. We extensively examine the spatial distribution of study sites, employed sensors, used datasets and CNN architectures, and give a thorough overview of applications in EO which used CNNs. Our main finding is that CNNs are in an advanced transition phase from computer vision to EO. Upon this, we argue that in the near future, investigations which analyze object dynamics with CNNs will have a significant impact on EO research. With a focus on EO applications in this Part II, we complete the methodological review provided in Part I.

scholarly journals Object Detection and Image Segmentation with Deep Learning on Earth Observation Data: A Review-Part I: Evolution and Recent Trends

2020 ◽  
Vol 12 (10) ◽  
pp. 1667 ◽  
Author(s):  
Thorsten Hoeser ◽  
Claudia Kuenzer
Keyword(s):  
Image Segmentation ◽  
Deep Learning ◽  
Object Detection ◽  
Great Influence ◽  
Adaptive Method ◽  
Earth Observation ◽  
Future Research ◽  
Observation Data ◽  
Theoretical Concepts ◽  
Earth Observation Data

Deep learning (DL) has great influence on large parts of science and increasingly established itself as an adaptive method for new challenges in the field of Earth observation (EO). Nevertheless, the entry barriers for EO researchers are high due to the dense and rapidly developing field mainly driven by advances in computer vision (CV). To lower the barriers for researchers in EO, this review gives an overview of the evolution of DL with a focus on image segmentation and object detection in convolutional neural networks (CNN). The survey starts in 2012, when a CNN set new standards in image recognition, and lasts until late 2019. Thereby, we highlight the connections between the most important CNN architectures and cornerstones coming from CV in order to alleviate the evaluation of modern DL models. Furthermore, we briefly outline the evolution of the most popular DL frameworks and provide a summary of datasets in EO. By discussing well performing DL architectures on these datasets as well as reflecting on advances made in CV and their impact on future research in EO, we narrow the gap between the reviewed, theoretical concepts from CV and practical application in EO.

scholarly journals Quantifying Tree Cover Loss in Urban Forests within Nairobi City Metropolitan Area from Earth Observation Data

2020 ◽  
Vol 3 (1) ◽  
pp. 78
Author(s):  
Francis Oloo ◽  
Godwin Murithi ◽  
Charlynne Jepkosgei
Keyword(s):  
Land Cover ◽  
Metropolitan Area ◽  
Land Surface ◽  
Urban Forest ◽  
Urban Forests ◽  
Earth Observation ◽  
Tree Cover ◽  
Observation Data ◽  
Vegetation Health ◽  
Earth Observation Data

Urban forests contribute significantly to the ecological integrity of urban areas and the quality of life of urban dwellers through air quality control, energy conservation, improving urban hydrology, and regulation of land surface temperatures (LST). However, urban forests are under threat due to human activities, natural calamities, and bioinvasion continually decimating forest cover. Few studies have used fine-scaled Earth observation data to understand the dynamics of tree cover loss in urban forests and the sustainability of such forests in the face of increasing urban population. The aim of this work was to quantify the spatial and temporal changes in urban forest characteristics and to assess the potential drivers of such changes. We used data on tree cover, normalized difference vegetation index (NDVI), and land cover change to quantify tree cover loss and changes in vegetation health in urban forests within the Nairobi metropolitan area in Kenya. We also used land cover data to visualize the potential link between tree cover loss and changes in land use characteristics. From approximately 6600 hectares (ha) of forest land, 720 ha have been lost between 2000 and 2019, representing about 11% loss in 20 years. In six of the urban forests, the trend of loss was positive, indicating a continuing disturbance of urban forests around Nairobi. Conversely, there was a negative trend in the annual mean NDVI values for each of the forests, indicating a potential deterioration of the vegetation health in the forests. A preliminary, visual inspection of high-resolution imagery in sample areas of tree cover loss showed that the main drivers of loss are the conversion of forest lands to residential areas and farmlands, implementation of big infrastructure projects that pass through the forests, and extraction of timber and other resources to support urban developments. The outcome of this study reveals the value of Earth observation data in monitoring urban forest resources.

scholarly journals Reservoir-Induced Land Deformation: Case Study from the Grand Ethiopian Renaissance Dam

2021 ◽  
Vol 13 (5) ◽  
pp. 874
Author(s):  
Yu Chen ◽  
Mohamed Ahmed ◽  
Natthachet Tangdamrongsub ◽  
Dorina Murgulet
Keyword(s):  
Land Surface ◽  
Large Scale ◽  
Surface Deformation ◽  
Vertical Displacement ◽  
Nile River ◽  
Reservoir Volume ◽  
Novel Approach ◽  
Land Deformation ◽  
Large Scale Land ◽  
The Impact

The Nile River stretches from south to north throughout the Nile River Basin (NRB) in Northeast Africa. Ethiopia, where the Blue Nile originates, has begun the construction of the Grand Ethiopian Renaissance Dam (GERD), which will be used to generate electricity. However, the impact of the GERD on land deformation caused by significant water relocation has not been rigorously considered in the scientific research. In this study, we develop a novel approach for predicting large-scale land deformation induced by the construction of the GERD reservoir. We also investigate the limitations of using the Gravity Recovery and Climate Experiment Follow On (GRACE-FO) mission to detect GERD-induced land deformation. We simulated three land deformation scenarios related to filling the expected reservoir volume, 70 km3, using 5-, 10-, and 15-year filling scenarios. The results indicated: (i) trends in downward vertical displacement estimated at −17.79 ± 0.02, −8.90 ± 0.09, and −5.94 ± 0.05 mm/year, for the 5-, 10-, and 15-year filling scenarios, respectively; (ii) the western (eastern) parts of the GERD reservoir are estimated to move toward the reservoir’s center by +0.98 ± 0.01 (−0.98 ± 0.01), +0.48 ± 0.00 (−0.48 ± 0.00), and +0.33 ± 0.00 (−0.33 ± 0.00) mm/year, under the 5-, 10- and 15-year filling strategies, respectively; (iii) the northern part of the GERD reservoir is moving southward by +1.28 ± 0.02, +0.64 ± 0.01, and +0.43 ± 0.00 mm/year, while the southern part is moving northward by −3.75 ± 0.04, −1.87 ± 0.02, and −1.25 ± 0.01 mm/year, during the three examined scenarios, respectively; and (iv) the GRACE-FO mission can only detect 15% of the large-scale land deformation produced by the GERD reservoir. Methods and results demonstrated in this study provide insights into possible impacts of reservoir impoundment on land surface deformation, which can be adopted into the GERD project or similar future dam construction plans.

Social Distancing Using AI and Deep Learning

2021 ◽  
Vol 9 (VII) ◽  
pp. 1816-1822
Author(s):  
Mr. Kiran Mudaraddi
Keyword(s):  
Deep Learning ◽  
Object Detection ◽  
Open Source ◽  
Suggested Method ◽  
Video Frame ◽  
Social Distancing ◽  
The Social ◽  
The Impact

The paper presents a deep learning-based methodology for detecting social distancing in order to assess the distance between people in order to mitigate the impact of the coronavirus pandemic. The input was a video frame from the camera, and the open-source object detection was pre-trained. The outcome demonstrates that the suggested method is capable of determining the social distancing measures between many participants in a video.

scholarly journals Continuous high-resolution midlatitude-belt simulations for July–August 2013 with WRF

2017 ◽  
Vol 10 (5) ◽  
pp. 2031-2055 ◽  
Author(s):  
Thomas Schwitalla ◽  
Hans-Stefan Bauer ◽  
Volker Wulfmeyer ◽  
Kirsten Warrach-Sagi
Keyword(s):  
High Resolution ◽  
Land Surface ◽  
Large Scale ◽  
Temperature Fields ◽  
Surface Model ◽  
Limited Area ◽  
Data Set ◽  
Latitude Belt ◽  
Simulated Precipitation ◽  
The Impact

Abstract. Increasing computational resources and the demands of impact modelers, stake holders, and society envision seasonal and climate simulations with the convection-permitting resolution. So far such a resolution is only achieved with a limited-area model whose results are impacted by zonal and meridional boundaries. Here, we present the setup of a latitude-belt domain that reduces disturbances originating from the western and eastern boundaries and therefore allows for studying the impact of model resolution and physical parameterization. The Weather Research and Forecasting (WRF) model coupled to the NOAH land–surface model was operated during July and August 2013 at two different horizontal resolutions, namely 0.03 (HIRES) and 0.12° (LOWRES). Both simulations were forced by the European Centre for Medium-Range Weather Forecasts (ECMWF) operational analysis data at the northern and southern domain boundaries, and the high-resolution Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) data at the sea surface.The simulations are compared to the operational ECMWF analysis for the representation of large-scale features. To analyze the simulated precipitation, the operational ECMWF forecast, the CPC MORPHing (CMORPH), and the ENSEMBLES gridded observation precipitation data set (E-OBS) were used as references.Analyzing pressure, geopotential height, wind, and temperature fields as well as precipitation revealed (1) a benefit from the higher resolution concerning the reduction of monthly biases, root mean square error, and an improved Pearson skill score, and (2) deficiencies in the physical parameterizations leading to notable biases in distinct regions like the polar Atlantic for the LOWRES simulation, the North Pacific, and Inner Mongolia for both resolutions.In summary, the application of a latitude belt on a convection-permitting resolution shows promising results that are beneficial for future seasonal forecasting.

A Stochastic Parameter Perturbation Method to Represent Uncertainty in a Microphysics Scheme

2021 ◽  
Author(s):  
Gregory Thompson ◽  
Judith Berner ◽  
Maria Frediani ◽  
Jason A. Otkin ◽  
Sarah M. Griffin
Keyword(s):  
Land Surface ◽  
Large Scale ◽  
Random Perturbation ◽  
Longwave Radiation ◽  
Parameter Uncertainties ◽  
Microphysics Scheme ◽  
Convective Systems ◽  
Temporal Correlations ◽  
Precipitation Characteristics ◽  
The Impact

AbstractCurrent state-of-the art regional numerical weather forecasts are run at horizontal grid spacings of a few kilometers, which permits medium to large-scale convective systems to be represented explicitly in the model. With the convection parameterization no longer active, much uncertainty in the formulation of subgrid-scale processes moves to other areas such as the cloud microphysical, turbulence, and land-surface parameterizations. The goal of this study is to investigate experiments with stochastically-perturbed parameters (SPP) within a microphysics parameterization and the model’s horizontal diffusion coefficients. To estimate the “true” uncertainty due to parameter uncertainty, the magnitudes of the perturbations are chosen as realistic as possible and not with purposeful intent of maximal forecast impact as some prior work has done. Spatial inhomogeneities and temporal persistence are represented using a random perturbation pattern with spatial and temporal correlations. The impact on the distributions of various hydrometeors, precipitation characteristics, and solar/longwave radiation are quantified for a winter and summer case. In terms of upscale error growth, the impact is relatively small and consists primarily of triggering atmospheric instabilities in convectively unstable regions. In addition, small in situ changes with potentially large socio-economic impacts are observed in the precipitation characteristics such as maximum hail size. Albeit the impact of introducing physically-based parameter uncertainties within the bounds of aerosol uncertainties is small, their influence on the solar and longwave radiation balances may still have important implications for global model simulations of future climate scenarios.

Analyzing large-scale Earth Observation data repositories made simple with OpenEO Platform

2021 ◽  
Author(s):  
Edzer Pebesma ◽  
Patrick Griffiths ◽  
Christian Briese ◽  
Alexander Jacob ◽  
Anze Skerlevaj ◽  
...  
Keyword(s):  
Large Scale ◽  
Virtual Machines ◽  
Google Earth ◽  
Earth Observation ◽  
Data Cube ◽  
Observation Data ◽  
Data Repositories ◽  
Software Ecosystem ◽  
Wide Range ◽  
Earth Observation Data

<p>The OpenEO API allows the analysis of large amounts of Earth Observation data using a high-level abstraction of data and processes. Rather than focusing on the management of virtual machines and millions of imagery files, it allows to create jobs that take a spatio-temporal section of an image collection (such as Sentinel L2A), and treat it as a data cube. Processes iterate or aggregate over pixels, spatial areas, spectral bands, or time series, while working at arbitrary spatial resolution. This pattern, pioneered by Google Earth Engine™ (GEE), lets the user focus on the science rather than on data management.</p><p>The openEO H2020 project (2017-2020) has developed the API as well as an ecosystem of software around it, including clients (JavaScript, Python, R, QGIS, browser-based), back-ends that translate API calls into existing image analysis or GIS software or services (for Sentinel Hub, WCPS, Open Data Cube, GRASS GIS, GeoTrellis/GeoPySpark, and GEE) as well as a hub that allows querying and searching openEO providers for their capabilities and datasets. The project demonstrated this software in a number of use cases, where identical processing instructions were sent to different implementations, allowing comparison of returned results.</p><p>A follow-up, ESA-funded project “openEO Platform” realizes the API and progresses the software ecosystem into operational services and applications that are accessible to everyone, that involve federated deployment (using the clouds managed by EODC, Terrascope, CreoDIAS and EuroDataCube), that will provide payment models (“pay per compute job”) conceived and implemented following the user community needs and that will use the EOSC (European Open Science Cloud) marketplace for dissemination and authentication. A wide range of large-scale cases studies will demonstrate the ability of the openEO Platform to scale to large data volumes.  The case studies to be addressed include on-demand ARD generation for SAR and multi-spectral data, agricultural demonstrators like crop type and condition monitoring, forestry services like near real time forest damage assessment as well as canopy cover mapping, environmental hazard monitoring of floods and air pollution as well as security applications in terms of vessel detection in the mediterranean sea.</p><p>While the landscape of cloud-based EO platforms and services has matured and diversified over the past decade, we believe there are strong advantages for scientists and government agencies to adopt the openEO approach. Beyond the absence of vendor/platform lock-in or EULA’s we mention the abilities to (i) run arbitrary user code (e.g. written in R or Python) close to the data, (ii) carry out scientific computations on an entirely open source software stack, (iii) integrate different platforms (e.g., different cloud providers offering different datasets), and (iv) help create and extend this software ecosystem. openEO uses the OpenAPI standard, aligns with modern OGC API standards, and uses the STAC (SpatioTemporal Asset Catalog) to describe image collections and image tiles.</p>

Deep Learning Based Active Monitoring for Anti-collision between Vessels and Bridges

2019 ◽  
Author(s):  
Limu Chen ◽  
Ye Xia ◽  
Dexiong Pan ◽  
Chengbin Wang
Keyword(s):  
Decision Making ◽  
Deep Learning ◽  
Object Detection ◽  
Large Scale ◽  
Data Augmentation ◽  
Information Support ◽  
Single Shot ◽  
Active Monitoring ◽  
Detection Model ◽  
Comparison Results

<p>Deep-learning based navigational object detection is discussed with respect to active monitoring system for anti-collision between vessel and bridge. Motion based object detection method widely used in existing anti-collision monitoring systems is incompetent in dealing with complicated and changeable waterway for its limitations in accuracy, robustness and efficiency. The video surveillance system proposed contains six modules, including image acquisition, detection, tracking, prediction, risk evaluation and decision-making, and the detection module is discussed in detail. A vessel-exclusive dataset with tons of image samples is established for neural network training and a SSD (Single Shot MultiBox Detector) based object detection model with both universality and pertinence is generated attributing to tactics of sample filtering, data augmentation and large-scale optimization, which make it capable of stable and intelligent vessel detection. Comparison results with conventional methods indicate that the proposed deep-learning method shows remarkable advantages in robustness, accuracy, efficiency and intelligence. In-situ test is carried out at Songpu Bridge in Shanghai, and the results illustrate that the method is qualified for long-term monitoring and providing information support for further analysis and decision making.</p>

Sign in / Sign up

Export Citation Format

Share Document