Dataset

Road Surface Data
This dataset provides detailed information on road surfaces, distinguishing between paved and unpaved surfaces across various regions. It is derived from OpenStreetMap (OSM) data and enhanced with predictions from a hybrid deep learning approach. The data is augmented using images from Mapillary, the GHSU Global Human Settlement Urban Layer 2019, and the AFRICAPOLIS2020 urban layer.

Read more

This dataset supports the analysis and figures presented in the Global Urban OSM Building Completeness Analysis manuscript. It includes OSM completeness values and updated covariates for 13,189 urban agglomerations worldwide. The analysis employs a machine-learning model to infer the completeness of OSM building stock data, addressing the uneven spatial coverage of existing data.

Read more

Paper

License

Reference date

Last updated on 2024-08-23

This dataset introduces a fine-tuned model for detecting Municipal Solid Waste (MSW) using UAV imagery. The model was evaluated in the Msimbazi delta in Dar es Salaam, Tanzania, achieving an F1 score of 0.92. The generated MSW pile map revealed significantly higher contamination levels in the Msimbazi River bed compared to surrounding areas.

Read more

Licence

Reference date

Last updated on 2023-05

This dataset focuses on water tank detection to support vector control of infectious diseases transmitted by Aedes Aegypti. Using a semi-supervised self-training approach based on open satellite imagery, the study mapped water tank density in Rio de Janeiro. A negative binomial generalized linear regression model evaluated the statistical significance of water tank density for modeling Aedes Aegypti distribution from 2019 to 2021. The semi-supervised model outperformed a supervised model by 22% in F1-score.

Read more

Licence

Reference date

Last updated on 2024-09

MapSwipe utilizes volunteer contributions to improve geospatial data by identifying infrastructure, monitoring environmental changes, and validating existing map data. Volunteers engage in tasks such as locating specific features, detecting structural or environmental modifications, and verifying data accuracy. The nature of the data collected varies by project, with each initiative having a distinct focus and data structure tailored to its objectives.

Read more

License

OSMLanduse data is a scientific dataset generated within the scope of the Horizon 2020 – LandSense project. It is a classification of Sentinel-2 imagery using a deep learning model trained on OSM landuse and landcover features. The data might contain errorneous classifications. The classification values in the raster dataset correspond to a subset of the well-known CORINE LandCover Classification.

Read more

Paper

coming soon

License

Reference date

Based on Sentinel-2 composites from 2020