Tree Detection and Height From Shadow

Tree Detection and Height From Shadow finds trees location and their height in the input image. For each pixel, it returns the probability of it to be part of a tree, and computes its height based on the length of its adjacent shadow.

The block runs on images with ground sampling distance (GSD) of 0.55m or less. It takes as input Pleiades Display (recommended) or SPOT Display Data Blocks.

The blocks generates the following outputs:

1. A GeoTIFF heat map image containing the probability for each pixel to be part of a tree.
2. A shapefile containing the height for each detected tree (meters).

The solution uses deep learning techniques for computer vision to detect trees and shadows and physics equations to compute the height using sun and satellite position information.

Due to its complexity, this analysis block requires a significant running time. There are no restrictions on image dimensions, but the computing time will increase with the size of the image.

Approximation of the cost & performance parameters for the current block version (actual data depends on region and data block used):

Pléiades Display Area (sqkm)	Time	Data Credits	Infrastructure Credits	Tree-Height-Block Credits
1	3 min	390	38	1260
10	30 min	3900	380	12600
100	5 hours	39000	3800	126000
1000	50 hours	390000	38000	1125000

The use cases include Infrastructure Vegetation Risk Monitoring, Urban Planning & Construction.

Supported Workflows

This block can be used with:

• Pléiades Display (Streaming) -> Raster Tiling -> Tree Detection and Height From Shadow (recommended)
• SPOT Display (Streaming) -> Raster Tiling -> Tree Detection and Height From Shadow
• Pléiades Display (Download) -> DIMAP -> GeoTIFF Conversion -> Raster Tiling -> Tree Detection and Height From Shadow (recommended)
• SPOT Display (Download) -> DIMAP -> GeoTIFF Conversion -> Raster Tiling -> Tree Detection and Height From Shadow

Future data blocks integrations will come in next versions – please reach out with the data block you would like to use and Spacept will prioritize its development.

General Information	Description
Block Type	Processing for Tree Detection and Height calculation
Supported input data	Raster Tiled (512x512) GeoTIFF images. The image is expected to have Ground Sampling Distance (GSD) less than 0.55 m.
Output data format	GeoTIFF image for tree probability and GeoJSON for height measurements.
Resolution	Same resolution as the input data for the tree probability and less than 5 meters on the ground for the polygons for height measurements.
Performance	For tree height calculation, 1-2 meters accuracy on isolated objects (entire shadow on level ground), up to 8 meters in worse conditions.

Example Job Parameters

{
   "tiling:1":{
      "nodata":null,
      "tile_width":512,
      "tile_height":512,
      "match_extents":false,
      "output_prefix":"",
      "augmentation_factor":1,
      "discard_empty_tiles":true
   },
   "tree-height-detection:1":{
      
   },
   "oneatlas-pleiades-aoiclipped:1":{
      "ids":null,
      "bbox":[
         -1.039205,
         46.732618,
         -1.026172,
         46.743978
      ],
      "time":"2018-01-01T00:00:00+00:00/2020-12-31T23:59:59+00:00",
      "limit":1,
      "zoom_level":18,
      "time_series":null,
      "max_cloud_cover":100,
      "panchromatic_band":false
   }
}

More information

For more information about this block, please check the provider website.