Try Gemini 1.5 Pro, our most advanced multimodal model in Vertex AI, and see what you can build with a 1M token context window. Try Gemini 1.5 Pro, our most advanced multimodal model in Vertex AI, and see what you can build with a 1M token context window.

Create an annotation set

An annotation set is a set of labels that you apply to the data of a dataset for model training. Annotation sets are associated with a data type and model objective (for example, video/classification). You create annotation sets for image, text, and video data types.

To learn how to apply the labels of your annotation set to your dataset's data, see Label using the Google Cloud console.

How to create an annotation set

When you create a dataset in Vertex AI, if the imported CSV or JSON Lines file contains labels, Vertex AI automatically creates an annotation set and assigns it the objective that you selected for the dataset. For example, if you select Video as the data type and Classification as the objective, the annotation set DATASET_NAME\_vcn is created, where "vcn" stands for "video classification."

If the dataset is already created and you want to add an annotation set or additional annotation sets, you can create them in the Google Cloud console as detailed in the following section.

Create a new annotation set

You can create a new annotation set for an existing dataset in the Google Cloud console only.

In the Google Cloud console, go to the Vertex AI Datasets page.

Go to Datasets
Click the dataset that you want to create an annotation set for.

The dataset appears.
In the selector box next to the name of your dataset, select Create annotation set.
In the Create annotation set pane, enter a name for the annotation set.
Select the model objective.
Click Create.

The dataset view appears with your new annotation set displayed. You are now ready to apply the labels of your new annotation set to your data.

What's next

Edit, or add labels to your dataset using the Google Cloud console.