All Collections
Annotate Media
Create Categories for annotation
Create Categories for annotation

How to create training data specific to your needs

Paige Kassalen avatar
Written by Paige Kassalen
Updated over a week ago

In a Project, Categories are the individual objects, features, or characteristics you want to build a computer vision model to identify. So, for example, if you were building a model to differentiate cats vs. dogs in pictures, your Project would have two Categories: cat, dog.

(At CrowdAI, we call these Categories, but elsewhere you may hear the term classes or even ontology, which have similar meanings. We think Categories is a bit easier to understand, though.)

Create and View Categories in a Project

To create a new Category, first go to the Annotate page of the Project, and scroll down to the What's Being Annotated box. From here, you can see any Categories that you've already attached to this Project.

Click the Manage categories link to open the Category Manager.

Within the Category Manager, you can see the list of all Categories created within your account. This list includes all Categories across all Projects.

Click Add Category to create a new Category.

First, you'll want to give your Category a name. Keep it short, but clear and descriptive. Spaces are fine, but it's often a good idea to use hyphens or underscores to connect words in a Category name, e.g. flat-tire.

If what you're creating is actually a subcategory of another Category, use the Category Nesting field to select the "parent" Category this new one will fall under.

Next, select a label type and color. Each Category can have only one label type (e.g. polygon).

Last, you'll want to enter instructions for this particular Category. Use the rich text editor to add bullets, numbered lists, or even images to help illustrate how someone should annotate this Category when they come across it in your media!

Did this answer your question?