Installation

OneOpen ML Studio requires Python 3.10+ and a modern browser. GPU is optional but recommended for YOLO Label Assist, SAM, and training.

From PyPI (recommended)

pip install oneopen-ml-studio

Extra	Install command	Enables
`yolo`	`pip install "oneopen-ml-studio[yolo]"`	Ultralytics Label Assist + training
`libreyolo`	`pip install "oneopen-ml-studio[libreyolo]"`	LibreYOLO training backend
`sam`	`pip install "oneopen-ml-studio[sam]"`	PyTorch deps for SAM (see note below)
`detectors`	`pip install "oneopen-ml-studio[detectors]"`	Torchvision detect / cls / seg catalogs
`face`	`pip install "oneopen-ml-studio[face]"`	Face-tagged Torchvision extras
`tabular`	`pip install "oneopen-ml-studio[tabular]"`	scikit-learn tabular / text train
`nlp`	`pip install "oneopen-ml-studio[nlp]"`	Hugging Face prepare helpers
`server`	`pip install "oneopen-ml-studio[server]"`	Postgres, Redis, Celery, boto3 (Team)
`all`	`pip install "oneopen-ml-studio[all]"`	yolo + sam + libreyolo + detectors + face + tabular + nlp
`docs`	`pip install "oneopen-ml-studio[docs]"`	Sphinx docs build tools
`dev`	`pip install "oneopen-ml-studio[dev]"`	pytest, ruff

Tip

For most users starting out, install with YOLO extras:

pip install "oneopen-ml-studio[yolo]"

Note

The [sam] extra installs torch and torchvision only. You still need the Meta segment-anything package and a checkpoint file. See SAM (interactive segmentation).

git clone https://github.com/1-OpenSource/OneOpen-ML-Studio.git
cd OneOpen-ML-Studio
pip install -e ".[yolo]"

Editable install is useful when contributing or hacking on the UI/templates.

oneopen version
# OneOpen ML Studio v1.0.0

Or:

python -m oneopen_ml_studio version

Platform	Notes
Windows	Fully supported. Training uses `workers=0` for DataLoaders to avoid spawn crashes.
Linux / macOS	Fully supported. Training defaults to 4 DataLoader workers.
GPU	CUDA via PyTorch/Ultralytics when available. Device can be set to Auto / GPU / CPU in the Train UI.

pip install "oneopen-ml-studio[docs]"
# or: pip install -r docs/requirements.txt
cd docs
make html          # Unix
.\make.bat html    # Windows

Output is static HTML in docs/_build/html/. Open index.html locally, or deploy that folder to your host.

Published docs: https://oneopensource.org/oneopen-ml-studio/

Preferred local command: oneopen start (alias: oneopen serve).