luxonis-ml 0.8.2

MLOps tools for training models for Luxonis devices

LuxonisML

This library includes a collection of helper functions and utilities for the Luxonis MLOps stack. This includes the following submodules:

• Dataset Management: Creating computer vision datasets focused around Luxonis hardware and to be used with our LuxonisTrain framework. Additional documentation can be found here.
• Tracking: Our implementation of a logger for use with PyTorch Lightning or in LuxonisTrain
• Utils: Miscellaneous utils for developers.

NOTE: The project is in a beta state, it might be missing certain features or contain bugs - please report any feedback!

Table of Contents

• Installation
• CLI
• Contributing

Installation

The luxonis_ml package is hosted on PyPI, so you can install it with pip.

We offer several versions of the package:

• luxonis-ml[data]: installs necessary dependencies for using luxonis_ml.data module
• luxonis-ml[utils]: installs necessary dependencies for using luxonis_ml.utils module
• luxonis-ml[tracker]: installs necessary dependencies for using luxonis_ml.tracker module
• luxonis-ml[all]: installs all dependencies
• luxonis-ml[dev]: installs all dependencies, including development dependencies

Additional dependencies

Additional dependencies for working with specific cloud services can be installed using the following extras:

• gcs: Dependencies for working with Google Cloud Storage
• s3: Dependencies for working with AWS S3
• roboflow: Dependencies for downloading datasets from Roboflow
• mlflow: Dependencies for working with MLFlow

[!NOTE] If some of the additional dependencies are required but not installed (e.g. attempting to use Google Cloud Storage without installing the gcs extra), then the missing dependencies will be installed automatically.

Example:

Installing the package with the data extra and dependencies for gcs and roboflow:

pip install luxonis-ml[data,gcs,roboflow]

Installing the package with all dependencies:

pip install luxonis-ml[all]

CLI

The luxonis-ml package comes with a CLI that can be used to interact with the library.

To see the available commands, run:

luxonis_ml --help

Contributing

If you want to contribute to this project, read the instructions in CONTRIBUTING.md