GeoAI Amazon Basin Secret Runway Detection Challenge by ITU
Helping Austria, Peru
and 1 other country
and 1 other country
- Austria
- Peru
- Italy
- Scroll to see more
2 000 CHF
Challenge completed ~1 year ago
Classification
Earth Observation
394 joined
26 active
Start
Aug 09, 24
Close
Nov 11, 24
Reveal
Nov 11, 24
About
- Satellite images: the participants should use free-accessible satellite data. It is recommended the use of Sentinel-1 and Sentinel-2 images, which are considered particularly suitable for this challenge, leveraging both SAR and optical technologies. The data pre-processing methodology are not limited.. The participants need to share the availability of the data they used and how the pre-processing is conducted. Sentinel data are freely available in a variety of platforms (e.g., Copernicus Open Access Hub: https://scihub.copernicus.eu/ ; ), including cloud computing platforms such as Google Earth Engine (https://earthengine.google.com/).
- UNODC will provide a limited number of training samples, as polygon shapefile delineating the contour line of the runways. To each sample runway the year of detection will be indicated in the attribute table of the shapefile. This will allow participants to select the more suitable images for training the model (clandestine runways are a dynamic fenomenon and do not remain active for long period, thus the temporal component must be taken into account). The training sample provided by UNODC is validated through visual interpretation of high resolution optical images.
- Additional training samples can be added by participants using open resources (e.g., https://plataforma.brasil.mapbiomas.org/pistas-de-pouso). However, it is recommended to visually analyse alternative datasets for quality against satellite images.
- UNODC will provide test AOIs over which to run the devised models for clandestine runways detection . For each AOI the year of analysis will be specified.
Join the largest network for
data scientists and AI builders
Status
© Zindi 2025