custom features, overview
Some functionalities are missing in the actual custom features feature. The only purpose of this issue is to summarize the missing functionalities by pointing opened issues. Feel free to complete the list below according to the progress of development or to add new ideas in order to expand the custom features possibilities.
-
#197 (closed) use exogenous data -
#206 be able to use custom features and scikit-learn -
#204 (closed) the use of Sentinel-1 sensor -
#197 (closed) access mask and data before gapfilling and iota2FeatureExtraction execution -
add running tests for all sensors -
create a library file of pre-existing functions for custom features (like ones proposed in #17 ) -
add functions listed in #17 -
allow to compute features on a N*N neighbourhood. This point requires a efficient split of input images (#466 (closed)) -
Provide a "bandMath like service". If users does not want to write python code, offer a way to express a per date features as a bandMath expression. -
#445 allow users to get coordinates -
#393 (closed) give parameters to custom feature from documentation file -
#438 (closed) catch custom feature errors before starting the chain
Edited by Arthur Vincent