A5009590473- Land Use and Ecosystem Services
- Remote-Sensing Image Classification
- Remote Sensing and Land Use
- Remote Sensing in Agriculture
- Regional Economics and Spatial Analysis
- Regional Development and Policy
- Impact of Light on Environment and Health
- Cultural Industries and Urban Development
- Geographic Information Systems Studies
- Urban Transport and Accessibility
- Advanced Image and Video Retrieval Techniques
- Global Urban Networks and Dynamics
- French Urban and Social Studies
- Automated Road and Building Extraction
- Rangeland Management and Livestock Ecology
- Soil and Land Suitability Analysis
- Spatial and Panel Data Analysis
- Urban and Rural Development Challenges
- Remote Sensing and LiDAR Applications
- Landslides and related hazards
- Migration, Policy, and Dickens Studies
- Regional resilience and development
- Transport and Economic Policies
- Historical Geography and Geographical Thought
- Climate Change, Adaptation, Migration
Université Libre de Bruxelles
1999-2024
Vrije Universiteit Brussel
2021
Terre des Hommes
1999
Formulae display:?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax order to improve their display. Uncheck the box turn off. This feature requires Javascript. Click on a formula zoom.
In this letter, the recently developed extreme gradient boosting (Xgboost) classifier is implemented in a very high resolution (VHR) object-based urban land use-land cover application. detail, we investigated sensitivity of Xgboost to various sample sizes, as well feature selection (FS) by applying standard technique, correlation-based FS. We compared with benchmark classifiers such random forest (RF) and support vector machines (SVMs). The methods are applied VHR imagery two sub-Saharan...
This study evaluates the impact of four feature selection (FS) algorithms in an object-based image analysis framework for very-high-resolution land use-land cover classification. The selected FS algorithms, correlation-based selection, mean decrease accuracy, random forest (RF) based recursive elimination, and variable using forest, were tested on extreme gradient boosting, support vector machine, K-nearest neighbor, RF, partitioningclassifiers, respectively. results demonstrate that...
Up-to-date and reliable land-use information is essential for a variety of applications such as planning or monitoring the urban environment. This research presents workflow mapping land use at street block level, with focus on residential use, using very-high resolution satellite imagery derived land-cover maps input. We develop processing chain automated creation polygons from OpenStreetMap ancillary data. Spatial metrics other features are computed, followed by feature selection that...
This study presents the development of a semi-automated processing chain for urban object-based land-cover and land-use classification. The is implemented in Python relies on existing open-source software GRASS GIS R. complete tool available open access adaptable to specific user needs. For automation purposes, we developed two add-ons enabling users (1) optimize segmentation parameters an unsupervised manner (2) classify remote sensing data using several individual machine learning...
Land cover Classified maps obtained from deep learning methods such as Convolutional neural networks (CNNs) and fully convolutional (FCNs) usually have high classification accuracy but with the detailed structures of objects lost or smoothed. In this work, we develop a methodology based on (FCN) that is trained in an end-to-end fashion using aerial RGB images only input. Skip connections are introduced into FCN architecture to recover spatial details lower layers. The experiments conducted...
Earth observation (EO) capabilities to produce up-to-date geographical information on slums over large areas supporting urban planning and evidence-based policymaking are largely acknowledged. Most EO studies typically use a data-driven approach without an understanding of end-user requirements. This study addresses this gap by aligning methods with societal needs concerns using user-driven in Accra, Ghana. By carrying out in-situ observations slum experts interviews, we produced map that...
To classify Very-High-Resolution (VHR) imagery, Geographic Object Based Image Analysis (GEOBIA) is the most popular method used to produce high quality Land-Use/Land-Cover maps. A crucial step in GEOBIA appropriate parametrization of segmentation algorithm prior classification. However, little effort has been made automatically optimize algorithms an unsupervised and spatially meaningful manner. So far, Unsupervised Segmentation Parameter Optimization (USPO) techniques, assume spatial...
Built-up layers derived from medium resolution (MR) satellite information have proven their contribution to dasymetric mapping, but suffer important limitations when working at the intra-urban level, mainly due difficulty in capturing whole range of variation terms built-up densities. In this regard, very-high (VHR) remote sensing is known for its ability better capture small variations densities and derive detailed urban land use, which plead favor use mapping populations. paper, we compare...
Malaria burden is increasing in sub-Saharan cities because of rapid and uncontrolled urbanization.Yet very few studies have studied the interactions between urban environments malaria.Additionally, no standardized land-use/land-cover has been defined for malaria studies.Here, we demonstrate potential local climate zones (LCZs) modeling prevalence rate (Pf PR 2-10 ) studying settings across nine African cities.Using a random forest classification algorithm over set 365 surveys we: (i)...
Multitemporal environmental and urban studies are essential to guide policy making ultimately improve human wellbeing in the Global South. Land-cover products derived from historical aerial orthomosaics acquired decades ago can provide important evidence inform long-term studies. To reduce manual labelling effort by experts scale large, meaningful regions, we investigate this study how domain adaptation techniques deep learning help efficiently map land cover Central Africa. We propose...
In object-based image analysis (OBIA), the appropriate parametrization of segmentation algorithms is crucial for obtaining satisfactory classification results. One ways this can be done by unsupervised parameter optimization (USPO). A popular USPO method does through a “global score” (GS), which minimizes intrasegment heterogeneity and maximizes intersegment heterogeneity. However, calculated GS values are sensitive to minimum maximum ranges candidate segmentations. Previous research...
Abstract Background The rapid and often uncontrolled rural–urban migration in Sub-Saharan Africa is transforming urban landscapes expected to provide shelter for more than 50% of Africa’s population by 2030. Consequently, the burden malaria increasingly affecting population, while socio-economic inequalities within settings are intensified. Few studies, relying mostly on moderate high resolution datasets standard predictive variables such as building vegetation density, have tackled topic...
A systematic and precise understanding of urban socio-economic spatial inequalities in developing regions is needed to address global sustainability goals. At the intra-urban scale, access detailed databases (i.e., a census) often difficult exercise. Geolocated surveys such as Demographic Health Surveys (DHS) are rich alternative source information but can be challenging interpolate at fine scale due their displacement, survey design lack very high-resolution (VHR) predictor variables these...
Mapping large heterogeneous urban areas using object-based image analysis (OBIA) remains challenging, especially with respect to the segmentation process. This could be explained both by complex arrangement of land-cover classes and high diversity patterns which can encountered throughout scene. In this context, a single parameter obtain satisfying results for whole scene impossible. Nonetheless, it is possible subdivide city into smaller local zones, rather homogeneous according their...
With rapid urbanization leading to the proliferation of deprived urban areas (often referred as "slums") in sub-Saharan Africa, there is a growing number city dwellers living inadequate housing conditions and being exposed multiple hazards. In this context, Earth Observation has potential for filling gaps spatial data availability thereby support evidence-based policy making. We assess free open-source software, open dual-pol SAR optical imagery (Sentinel-l Sentinel-2), global datasets...
In this paper, we explore the hypothesis that contemporary geographies of economic activities within metropolitan areas in Europe are moving into polycentric configurations. Our contribution is based on results derived from an analysis 7 metropolises were brought together EU-funded research programme "COMET" (namely Amsterdam, Barcelona, Berlin, Brussels, Copenhagen, Strasbourg and Vienna). We first build enquiry enterprises these cities, then document nature localisation multiple...
Encouraged by the EU INSPIRE directive requirements and recommendations, Walloon authorities, similar to other regional or national want develop operational land-cover (LC) land-use (LU) mapping methods using existing geodata. Urban planners environmental monitoring stakeholders of Wallonia have rely on outdated, mixed, incomplete LC LU information. The current reference map is 10-years old. two object-based classification methods, i.e., a rule- classifier-based method, for detailed urban...
Literature on world cities has repeatedly argued that major agents of globalization, i.e. 'global players' such as advanced producer service firms, transnational corporations and international organisations, play a determining role in interlocking into networks control centres the contemporary economy. It is very rare however analyses cover spatial distribution these global players within urban settings. We argue here local geographies shed new light how globalization territorialised cities....