User guide

Installation

Metrics Petri requires Python 3.10 or later.

python -m venv .venv
source .venv/bin/activate
pip install metrics-petri

For a development checkout, install the package in editable mode:

pip install -e .

Analyse a folder

metrics-petri input_images/

The command writes a ZIP archive containing tabular results, provenance, overlays, and growth charts. Common options include:

metrics-petri input_images/ --output results/run01.zip
metrics-petri input_images/ --metadata input_images/image_metadata.csv
metrics-petri input_images/ --threshold 0.45
metrics-petri input_images/ --dish-size-mm 60
metrics-petri input_images/ --seed 123

Supplying metadata enables day-coded charts and growth-rate calculations. The batch pipeline guide describes the metadata workflow and output columns.

Physical measurements default to a 90 mm outside dish diameter. Use --dish-size-mm with the actual diameter when analysing another dish size. The PyTorch seed defaults to 0 and is recorded in provenance.json; pass --seed to document a different reproducibility setting.

Example dataset

The source repository includes three example images in input_images/06FEB for smoke-testing and tutorials:

metrics-petri input_images/06FEB --output /tmp/metrics-petri-06feb.zip --seed 0

Open the ZIP to inspect analysis_full.csv, analysis_full.json, provenance.json, per-image overlays, and growth charts.

Crop multi-dish photographs

metrics-petri-crop -i input_images/
metrics-petri-crop -i photo.jpg --date 06/Feb
metrics-petri-crop -i input_images/ --debug

The cropper saves fully visible dishes as individual images in a cropped/ directory unless another output location is supplied.

Model use

The packaged checkpoint segments fungal colony area before morphometric analysis. Read the model card for its training provenance, validation results, limitations, and out-of-scope uses.

Further information

The repository README includes the full workflow, notebook instructions, Make targets, measured features, and citation details.