ellipsisAI 0.1.0

Package to use Ellipsis Drive for AI

This package is meant to help you run models on Ellipsis Drive content. It should be used in combination with the ellipsis package.

Install

pip install ellipsisAI

Example

import ellipsisAI as ai
import ellipsis as el

blockId = '170aadad-8eaa-4509-9c0e-c1536d58a1fe'
captureId = "633b4b9f-d939-4c4a-8d90-0e9fceb64b83"
targetBlockId = "066458f4-f018-4f49-a1f0-dedfa71b3368"
temp_folder = 'YOUR_PATH'

#login to get a authentication token
token = el.logIn('YOUR_USERNAME','YOUR_PASSWORD')

#retrieve the zoom and bounds of the capture you wish to classify
classificationZoom = ai.getZoom(blockId, captureId, token)
bounds = ai.getBounds(blockId, captureId, token)


#we create a dummy model. We use the identity function mapping an image to itself. We use the getTleData function to retirve the image for the given input tile ofthe model.
def model(tile):
    image = ai.getTileData(blockId, captureId, tile, token)
    return(image)

#apply the model on the given bounds on the given zoomlevel
ai.applyModel(model, bounds, targetBlockId, classificationZoom, token, temp_folder)

Functions

getZoom

getZoom(blockId, captureId, token)

This function retrieves the max zoomlevel of the specified capture. The result can be used as classificationgZoom argument in the applyModel or getTiles function.

Name	Description
blockId	The id of the block the capture is in
captureId	The id of the capture
token	Your token

getBounds

getBounds(blockId, captureId, token)

This function retrieves the bounds of the specified capture. The result can be used as bounds argument in the applyModel function.

Name	Description
blockId	The id of the block the capture is in
captureId	The id of the capture
token	Your token

applyModel

applyModel(model, bounds, targetBlockId, classificationZoom, token, temp_folder, visualizationId = None, targetNoDataValue = 0, targetStartDate = None, targetEndDate = None)

This function applies the given model on all tiles of zoomlevel classificationZoom withing the specified bounds. The results will be written in a new capture of the specified target block.

Name	Description
model	A function mapping mapping a tile object to a 3 dimensional numpy array. The dimensions of the numpy array should be independent of the tile id. A tile object is a dictionary with keys 'tileX', 'tileY' and 'zoom' as integers.
bounds	A shapely polygon or multipolygon indicating the region you wish to classify
targetBlockId	The id of the block to write the result to
classificationZoom	The zoomlevel of the tiles you wish to use for the model input as integer.
token	Your token
temp_folder	A path where temporary files can be written
visualizationId	visualization to use as input, if not given original raster is used
targetNoDataValue	Which number of the model output to interpret as transparent
targetStartDate	datetime object with date to use for the capture to which the results will be written. Defaults to current date
targetEndDate	datetime object with date to use for the capture to which the results will be written. Defaults to targetStartDate

getTileData

getTileData(blockId, captureId, tile, token, visualizationId = None, downsampleFactor = 1 )

This function retrieves data for the given tile as numpy array. This function can be used within your model function te retireve relevant data for the given tile.

Name	Description
blockId	the id of the block to get data from
captureId	the id of the capture to get data from
tile	A dictionary with keys tileX, tileY and zoom as int
token	Your token
visualizationId	The id of the visualization you wish to retrieve. If not given raw data is used
downsampleFactor	A factor with which to downsample the data. Default 1 (no downsampling)

getTiles

getTiles(bounds, classificationZoom)

This function covers a given bounds with tiles of the given zoomlevel. You can use the result to get tile arguments for the getTileData function

Name	Description
bounds	A shapely polygon or multipolygon
classificationZoom	The zoomlevel of the tiles to cover with as int