How to create a project

Before you can begin labeling, you must first create a project. Each project is a training data management environment where you can manage the performance, quality, and progress of your labeling project. If you would rather do this programmatically, you can create a project via the Python SDK or the GraphQL API.

  1. Add new project
    The first step is to add a new project which is located in the Projects tab.
  1. Name your project
    Add a name to your project. You may optionally add a project description as well.
  1. Attach a dataset
    You can attach an existing dataset that you have already added or add a new dataset.
  1. Customize your labeling interface
    Before you can customize your labeling interface, you must choose a label editor. Unless you need to label geospatial data, the Editor has all of the labeling and QA tools you need. Once you select a label editor you'll be able to customize your ontology. See About the Labelbox ontology to learn how to create a custom ontology for your project.
  1. Attach labeler instructions
    After you select "complete setup", click on the "instructions" tab in the left panel to attach instructions for your labeling team to see when they open the labeling interface. You can also do this by selecting a project and navigating to "Settings" > "Label editor" > "Edit" > "Instructions".
  1. Start labeling
    Once setup is complete it will bring you to the project overview page where you can start labeling. This is also where you manage your project.

Was this page helpful?

Intro to the project dashboard

Contact