There are a few things to rework for the feature extractors. First, the base UDF imports should be handled using meta-programming, and not being hardcoded in the feature_extractors.py file, as it will inevitably causes conflicts later.
Our very first user also points that the output_labels function should be more dynamic. In his example, he wants to implement a custom Vegetation Index Calculator that requires user parameters and that will ultimately influence the output labels, which need to be set up statically in the processing graph after the UDF. One idea of implementing this would be to be able to optionally pass the user parameters directly in the output_labels parameters, so that it can access this information when running locally on GFMAP
[x] Make imports dynamic using meta-programming
[x] Allow the output_labels to be dynamic and depend on user parameters when running locally
There are a few things to rework for the feature extractors. First, the base UDF imports should be handled using meta-programming, and not being hardcoded in the feature_extractors.py file, as it will inevitably causes conflicts later.
Our very first user also points that the
output_labels
function should be more dynamic. In his example, he wants to implement a custom Vegetation Index Calculator that requires user parameters and that will ultimately influence the output labels, which need to be set up statically in the processing graph after the UDF. One idea of implementing this would be to be able to optionally pass the user parameters directly in theoutput_labels
parameters, so that it can access this information when running locally on GFMAP