Dataset enrichment in Visual Layer is a two-step process:

  1. Select one or more enrichment models
  2. Run enrichment to generate new metadata for your dataset

This process adds structured metadata (e.g., labels, captions) to help improve dataset quality, enable better filtering, and power downstream automation.


Prerequisites

Verify Dataset is Ready

To ensure your dataset is ready for enrichment:

  1. Navigate to the Dataset View and confirm the status is READY.
  2. Alternatively, check the dataset status in the Dataset Inventory.

Step 1: Select Models

1

Open the Enrich Tab

Go to your dataset and click the Enrich tab.

2

Choose a Model

Select a model from the list and click Generate Models Preview.

3

(Optional) Add More Models

Click Add Models to include additional enrichment models.

4

Preview Model Outputs

Switch between selected models to preview their metadata output before enrichment.


Step 2: Enrich the Dataset

1

Start Enrichment

Click Enrich Dataset to run all selected models on the dataset.

  • All images or frames will be enriched.
  • Processing time varies based on dataset size.
2

Track Progress

You can monitor enrichment status in either location:

  • Enrich tab inside the dataset view
  • Notifications panel in the Dataset Inventory

You can continue exploring and interacting with your dataset while enrichment is in progress.


Explore the Enriched Dataset

1

Choose Exploration View

Open your dataset and select either:

  • Images view(default)
  • Objects view
2

Review Enrichment Metadata

Expand the VL Enrichment section in the right panel to view model-generated metadata.

3

Customize Cluster Card View

Choose which metadata to display directly on cluster cards:

  • Labels
  • Captions
  • User Tags(manually created by users — see User Tags doc)