aiod-utils 0.1

Utilities for AI OnDemand

AI OnDemand (AIoD) Utilities

A central package to unify helpful utilities for AI OnDemand that are useful/used across the Nextflow pipeline, Segment-Flow, and the Napari plugin. This primarily covers a centralisation of I/O and the implementation of RLE format.

Installation

Requires Python 3.11 or 3.12.

Using pip:

pip install aiod_utils

Using uv:

uv add aiod_utils

For Bio-Formats support (e.g. .lif, .czi), install the optional extra:

pip install "aiod_utils[bioformats]"

What's included

• aiod_utils.io — Load images via BioIO, with automatic reader selection for common formats (TIFF, OME-TIFF, Zarr, ND2, and more).
• aiod_utils.rle — Encode and decode segmentation masks (binary and instance) as COCO-compatible Run-Length Encoding, with save/load support.
• aiod_utils.stacks — Utilities for splitting large volumetric images into memory-bounded substacks for use in our Nextflow pipeline (Segment-Flow).
• aiod_utils.preprocess — Modular image preprocessing steps (e.g. CLAHE, downsampling) with a base class for defining custom steps. Easily extendable for use in Segment-Flow or our Napari plugin.

Documentation

For the wider AIoD documentation, please see our docs.

License

MIT — see LICENSE.