Dataset Snapshots

Create, restore, and clone named checkpoints of your dataset state for version control and safe rollbacks.

Snapshots provide version control for your datasets. Each snapshot captures the exact state of a dataset — media, metadata, enrichment models, clustering, tags, and labels — at a specific point in time. Every dataset starts with an automatically created snapshot, and each pipeline operation (dataset creation, media addition, re-indexing) automatically creates a new revision before it runs. Restore any snapshot to roll back changes, or clone one to create an independent copy for parallel work.

Use snapshots to:

• Save a checkpoint before re-indexing, enrichment, or bulk metadata changes.
• Roll back to a known-good state after an operation produces unexpected results.
• Clone a dataset version to test different enrichment models without affecting the original.

Snapshot Create, Restore, and Clone actions require the Editor or Owner dataset role. Viewers cannot perform these operations.

Snapshot Create, Restore, and Clone are disabled while other tasks (adding media, enrichment, re-indexing) are running. Wait for in-progress operations to complete before working with snapshots.

Accessing the Snapshot Catalog

The Snapshot Catalog lists all saved snapshots for a dataset in chronological order.

1. Navigate to the dataset you want to manage.
2. Click the Snapshots icon in the dataset header toolbar.

The Snapshot Catalog drawer opens on the right side of the screen.

Snapshot Catalog Fields

Each snapshot entry in the catalog displays the following information:

Field	Description
Name	The snapshot name. Editable after creation.
Status Badge	Displays Current Version for the active snapshot, or Restoring during a restore operation.
Description	Optional description text, if provided at creation. Editable after creation.
Date and Time	When the snapshot was created.
Items	Total number of images and videos in the dataset at the time of the snapshot.
Created by	The email address of the user who created the snapshot.
Three-Dot Menu	Access to Edit, Clone, Copy Snapshot ID, Restore, and Remove actions.

Use the search bar at the top of the catalog to filter snapshots by name or description.

Creating a Snapshot

Create a snapshot to save the current state of your dataset before making changes. Automatic revisions are created by the system before each pipeline operation and do not require user action — this procedure describes creating a snapshot manually.

1. Open the Snapshot Catalog by clicking the Snapshots icon in the dataset header.
2. Click Create a snapshot.
3. Enter a Name for the snapshot.
4. Optionally, enter a Description to record why you created this checkpoint.
5. Click Create.

The snapshot is saved and appears at the top of the Snapshot Catalog. While the snapshot is being created, the Snapshots icon displays a spinning indicator.

If snapshot creation fails, a toast notification appears with the error details. The dataset returns to Ready status if the failure occurred in a recoverable stage. If the failure is unrecoverable, the dataset enters Error status.

Snapshots can only be created when the dataset is in Ready status. If the dataset is processing another operation, wait for it to complete before creating a snapshot.

Restoring a Snapshot

Restore a snapshot to revert the dataset to a previously saved state. All changes made since that snapshot — including media additions, enrichment, metadata updates, and tag changes — are replaced with the data saved in that snapshot. Restoration is available for any snapshot except the one marked as Current Version.

1. Open the Snapshot Catalog.
2. Locate the snapshot you want to restore.
3. Click the three-dot menu on the snapshot entry.
4. Select Restore.
5. In the confirmation dialog, click Yes to confirm the restore operation.

The dataset enters Read Only status while the restore processes. A Snapshot Restore task appears in the Task Manager where you can track progress. The dataset returns to Ready status once the restore completes.

Restore replaces all current dataset data with the saved state of the snapshot. This operation cannot be undone. Create a new snapshot of the current state before restoring if you want to preserve it.

Cloning from a Snapshot

Clone a snapshot to create a new, independent dataset from a saved state. The original dataset is unaffected.

1. Open the Snapshot Catalog.
2. Locate the snapshot you want to clone.
3. Click the three-dot menu on the snapshot entry.
4. Select Clone.
5. Confirm the clone operation in the dialog that appears.

A new dataset is created with the name {Original Dataset Name} (cloned), and an initial snapshot is created in the new dataset. A Snapshot Clone task appears in the Task Manager where you can track progress. Restore, Clone, and Remove are disabled on the source snapshot while the operation is in progress. The cloned dataset appears in the Dataset Inventory once processing completes.

The cloned dataset is fully independent. Changes to the clone do not affect the original dataset, and changes to the original do not affect the clone.

Editing a Snapshot

Update the name or description of any snapshot without affecting its saved data.

1. Open the Snapshot Catalog.
2. Click the three-dot menu on the snapshot you want to edit.
3. Select Edit.
4. Update the Name or Description.
5. Click Update.

Removing a Snapshot

Remove a snapshot you no longer need from the catalog.

1. Open the Snapshot Catalog.
2. Click the three-dot menu on the snapshot you want to remove.
3. Select Remove.
4. In the confirmation dialog, click Yes to confirm the removal.

The snapshot is removed from the catalog. Removing a snapshot does not affect the dataset, its current data, or any datasets previously cloned from the snapshot.

The snapshot marked as Current Version cannot be removed.

Copying a Snapshot ID

Copy the unique identifier for a snapshot for use with the API or when contacting support.

1. Open the Snapshot Catalog.
2. Click the three-dot menu on the relevant snapshot.
3. Select Copy Snapshot ID.

The ID is copied to your clipboard.

Navigating Between Versions

The Snapshot Catalog allows restoration to any snapshot, whether it was created earlier or later than the current version. This enables movement both forward and backward through the dataset history — restoring to an older checkpoint or advancing to a more recent one.

Snapshot Create, Restore, and Clone operations put the dataset in Read Only status and block all data-modifying tasks until the operation completes.

Snapshot Limitations

Keep the following constraints in mind when working with snapshots:

• Creating a snapshot requires the dataset to be in Ready status. Restore and clone also accept Partial Index and Error statuses.
• Snapshot creation is blocked while another operation is running — enrichment, media addition, re-indexing, label propagation, custom metadata import, or another snapshot restore.
• Restore and clone operations put the dataset in Read Only status, blocking all data-modifying tasks until the operation completes.
• Clone operations run on the new dataset, so they do not block operations on the source dataset.
• The snapshot marked as Current Version cannot be deleted. Create a new snapshot to change the current version first.
• Snapshots are available only for datasets created with snapshot support enabled. Older datasets may not support this feature.

Related Resources

Snapshots API

Manage snapshots programmatically — create, list, restore, clone, and delete via the REST API.

Task Manager

Track snapshot restore and clone operations alongside other dataset tasks.

Export Dataset

Export a specific version of your dataset after restoring a snapshot.

Enrichment

Create a snapshot before running enrichment models as a safety net.