Spark Utilities for Label Data

Utilities for label multisets. Currently available:

• convert integer type n5 dataset into label multiset dataset
• downsample label multiset n5 dataset
• extract-unique-labels from a label dataset
• downsample-unique-labels create mip-map for unique labels
• label-to-block-mapping generate a mapping from label to containing blocks for a dataset of unique label lists

Compile

To compile for use on local machine, run

mvn -Denforcer.skip=true -PfatWithSpark clean package

To compile for deployment on the Janelia cluster, run

mvn -Denforcer.skip=true -Pfat clean package

Run

To run locally, run

java \
    -Dspark.master=local[*] \
    -jar target/label-utilities-spark-0.1.0-SNAPSHOT-shaded.jar COMMAND [--help/-h] [COMMAND ARG...]