Key Concepts

Learn how Workspaces, Projects, Versions, and Models structure your computer vision work in Roboflow

In Roboflow, your work is organized into Projects, which are always contained within Workspaces. These are the two primary structures you'll interact with.

• Workspaces: Workspaces are collaborative environments designed for you and your team members. They serve as the central hub for organizing all your Projects. Your Roboflow plan and billing are managed at the Workspace level.

• Projects: Each Project in Roboflow is dedicated to organizing your computer vision work around a single computer vision dataset. This is where you'll manage your images, annotations, and the evolution of that dataset through its dataset versions.

  • Versions: Versions are point-in-time snapshots of your project's dataset. You can create a version from the current state of your dataset at any time. Once created, a version remains constant. Any subsequent changes you make to your dataset (like adding/removing images or modifying annotations) will not affect it. This is key for model reproducibility and comparison.

  • Models: These are the computer vision end-products, either generated after you train within Roboflow or uploaded by you, and are what you can deploy and use.

    • When you train a model on a dataset version, your model will be associated with that specific version. If you upload a pre-trained model, you can choose to associate it with a dataset version in your project or designate it as 'versionless.' Learn more about how to train models on Roboflow or upload trained models to Roboflow.