Using YOLOv8 as the Backend for Label Studio

Introduction

There are several repositories that use YOLOv8 as the backend, but none of them worked, so I modified them to make them functional.

Before Using

• Copy your pre-trained model 'best.pt' to this folder.
• Create an empty '.env' folder.
• Modify 'docker-compose.yaml' to match your environment for LABEL_STUDIO_BASEURL and LABEL_STUDIO_API_TOKEN.

Quickstart

Build and start Machine Learning backend on http://localhost:9090

docker-compose up

Check if it works:

$ curl http://localhost:9090/health
{"status":"UP"}

Then connect running backend to Label Studio using Machine Learning settings.

Referenced Repositories

• https://github.com/35grain/label-studio-yolov8-backend
• https://github.com/seblful/label-studio-yolov8-backend