CLI¶
Overview¶
The morphoclass package provides a command line interface (CLI) that allows to train a deep learning model on morphological data, and to use a trained model for inference. Moreover it provides some basic explain AI (XAI) functionality that allows to reason about the models predictions.
The entry point for the CLI is called morphoclass and is automatically
installed together with the morphoclass package. You can test that
the installation worked correctly by running the following command:
$ deepmorphs -V
deepmorphs, version 0.2.8.dev66
All functionality is organised into subcommands
deepmorphs <sub-command> ....
There are some options to append to the command line:
Options:
-h, --help Show this message and exit.
-v, --verbose The logging verbosity: 0=WARNING (default), 1=INFO, 2=DEBUG
--log-file PATH Write a copy of all output to this file.
-V, --version Show the version and exit.
The most useful one is -v / -vv, which enables a lot of useful output.
For interactive sessions I’d recommend to always use at least -v, .e.g
$ deepmorphs -v <sub-command> ...
Subcommands¶
At the moment there are obsolete / duplicate commands due to ongoing refactoring.
Currently used - done refactoring / being refactored:
preprocess-datasetPreprocess a raw morphology dataset.organise-datasetRead a morphology dataset from a CSV file andplot-dataset-statsIngest a morphology dataset through a CSV file and…trainTrain a morphology classification model.evaluateEvaluate a trained checkpoint.performance-tableGenerate a summary report of the performance of…morphometricsRun the morphometrics subcommand.xaiExplain model predictions
To refactor:
outlier-detectionOutlier detectiontransfer-learningTransfer learning
Old / obsolete:
train-and-evaluate(got split intotrainandevaluate)performance-reportreplaced byevaluateandperformance-table
Please see the DVC section for more information on what the commands do.