Marketplace

Wind Turbines Detection

Detects Wind Turbines on SPOT imagery tiles of 256 x 256 pixels with a margin of 128 pixels at Mercator Zoom 17.

Description

Wind Turbine Detection detects wind turbines locations on SPOT imagery.

Disclaimer: This block version currently uses a BETA version of the algorithm.

Technical information

This block uses a deep learning model for the classification of wind turbines.

FeatureInfo
Input parametersTiled imagery in GeoTIFF of any size. Expected ground resolution at 1.20 m. (i.e. WebMercator projection)
Output formatA GeoJSON file with each detected turbine as a specific polygon.
Training datasetThe algorithm has been trained on a 40,000 object dataset created specifically for this purpose.

Performance

The algorithm is expected to achieve a high level of recall. It is still considered as a Beta version.

Use cases

There are numerous use cases where it is important to know where wind turbines are located, typically energy production forecast and aircraft traffic regulations. Some databases are available but they do not cover all locations and are not accurate. Extracting their location from satellite imagery enables getting a current and precise view of the location. Revisiting at a later date also enables monitoring the construction of new wind turbines.

More information

For more information about this processing block, please see the provider website

Capabilities

Input Capabilities

raster
custom
augmentation_factor
2
up42_standard
dtype
uint8
format
GTiff
sensor
SPOT
tile_width
256
tile_height
256

Output Capabilities

vector
up42_standard
format
GeoJSON
geometry_type
Polygon

End User License Agreement

View the End User License Agreement conditions in a new tab.