Auto Label with Workflows: Label data with any model or custom logic
PreviousAsset Library: Find and work with all your imagesNextMCP Server: Upload custom model weights from your AI assistant
Last updated
Was this helpful?
You can now auto-label images with a saved Workflow instead of a single foundation or trained model. In the Auto Label model dropdown, open the Workflows tab, select a workflow compatible with your project type, map its output classes to your project classes, and preview the results on sample images before starting the job.
Labeling with a Workflow brings strategies that a single model cannot express. Slice large images with SAHI to catch small objects, combine models with consensus voting or two-stage crop and classify pipelines, filter predictions with custom logic, or label with foundation models like Gemini, Claude, and GPT. Jobs run in the background and apply only the classes you map to your project.
Last updated
Was this helpful?
Was this helpful?