YOLO

Ultralytics YOLO:

• https://github.com/ultralytics/ultralytics
• https://docs.ultralytics.com/tasks/

Deployment

Classifier

1. Create a worker using the Arkindex frontend. You can set whatever name/slug/type. We'll use:
   • Name: YOLO Classifier
   • Slug: yolo-classifier
   • Type: classifier
2. Import a new version for that worker with the following settings:
   • Docker image reference (from this registry): registry.gitlab.teklia.com/workers/yolo:1.3.0
   • YAML configuration: Paste the contents of the arkindex/yolo-classifier.yml file found in this repository.

Segmenter

1. Create a worker using the Arkindex frontend. You can set whatever name/slug/type. We'll use:
   • Name: YOLO Segmenter
   • Slug: yolo-segmenter
   • Type: image-segmenter
2. Import a new version for that worker with the following settings:
   • Docker image reference (from this registry): registry.gitlab.teklia.com/workers/yolo:1.3.0
   • YAML configuration: Paste the contents of the arkindex/yolo-segmenter.yml file found in this repository.

Models

You can use freely available open-source models provided by Ultralytics:

1. Create a directory where you'll download a model: mkdir yolo-model
2. Download model weights using one of the links from the webpage above, and rename it model.pt:

curl -L https://github.com/ultralytics/assets/releases/download/v8.2.0/yolov8s-world.pt > yolo-model/model.pt

3. Write a .arkindex.yml with the following content, updating the name to match your model:

version: 2

models:
  - path: yolo-model
    name: YOLOv8s-Monde

4. Upload the model using Arkindex CLI: arkindex models publish
5. The model is now available on your Arkindex instance!

Development

For development and tests purpose it may be useful to install the worker as a editable package with pip.

pip install -e .

Linter

Code syntax is analyzed before submitting the code.
To run the linter tools suite you may use pre-commit.

pip install pre-commit
pre-commit run -a

Run tests

Tests are executed with tox using pytest.

Tests are executed with tox using pytest. We recommend installing it with tox-uv, like it is done in CI, to allow faster tests execution.

pip install tox-uv
uv tool install tox --with tox-uv
tox

To recreate tox virtual environment (e.g. a dependencies update), you may run tox -r