constantinpape / hdf-n5-converter

converter for turning labeled hdf5 data into n5 label dataset

HDF to N5 Converter

Command-line converter tool for turning labeled HDF5 data into an N5 label dataset of arbitrary cell size, which allows for parallel writing. In the future, BigCAT will probably use N5 as the backend for label data.

Compile

To compile a "fat jar" with all dependencies added, run:

mvn clean compile assembly:single

Usage

java -jar target/hdf-n5-converter-0.0.1-SNAPSHOT-jar-with-dependencies.jar [args]

Arguments

• --cellsize, -cs Size of cells to use in the output N5 dataset. Default: 64,64,8

• --compression, -c Compression type to use in output N5 dataset. Default: RAW

• --datasetname, --data, -d Output dataset name (N5 relative path from group) (Required)

• --groupname, --group, -g Output group name (N5 group) (Required)

• --inputfile, --input, -i Input HDF5 file (Required)

• --label, -l Path to labels in HDF5 file. Default:/volumes/labels/neuron_ids

Example

java -jar target/hdf-n5-converter-0.0.1-SNAPSHOT-jar-with-dependencies.jar --input ~/sample_A_20160501.hdf --group ~/n5 --data sampleA-fullres --compression GZIP

Would write an N5 dataset at ~/n5/sampleA-fullres from the labels at /volumes/labels/neuron_ids in ~/sample_A_20160501.hdf with GZIP compression (recommended).

---

NOTE: This currently requires LabelUtils from the n5cacheloader branch of BigCAT, which currently is not merged into master. To compile this project from source, you must check out that branch and maven install it so that it has the necessary functionality.