Manage Assets
Organize and maintain your dataset assets through deletion and bulk operations.
Effective asset management is essential for maintaining high-quality datasets. Datature Vi provides tools to delete individual assets and perform bulk operations, helping you clean up data, remove duplicates, and optimize your datasets for training.
Important considerations
- Deletion is permanent — Deleted assets cannot be recovered
- Annotations are removed — All annotations associated with deleted assets are permanently lost
- Always backup — Download your dataset before performing large-scale deletions
- Review carefully — Double-check selections before confirming deletions
Asset management operations
Remove individual assets one at a time using the asset action menu
Delete multiple assets simultaneously with selection and bulk delete
When to manage assets
Regular asset management helps maintain dataset quality and optimize storage:
| Scenario | Action | Best practice |
|---|---|---|
| Poor quality data | Delete individual assets | Review and remove blurry, corrupted, or unusable images |
| Duplicate content | Use bulk delete | Select and remove multiple copies at once |
| Wrong dataset | Delete individual or bulk | Remove assets uploaded to incorrect dataset |
| Dataset cleanup | Use bulk delete | Clear out test data or placeholder images |
| Storage optimization | Delete unused assets | Remove unnecessary data to reduce costs |
| Privacy compliance | Delete sensitive data | Remove assets containing private information |
Choosing the right approach
Delete individual assets
Best for:
- Removing specific low-quality images
- Selective cleanup during data review
- Removing individual problematic assets
- Precision deletion of specific files
Learn how to delete individual assets →
Use bulk operations
Best for:
- Large-scale dataset cleanup
- Removing multiple similar assets
- Clearing out entire categories of data
- Efficient batch deletion workflows
Learn how to use bulk actions →
Asset deletion workflow
Follow this recommended workflow for safe asset management:
1. Backup your dataset
Before any deletion operations, download a backup:
Explorer → Annotations → Export Annotations2. Identify assets to remove
Review your assets and determine which ones need deletion:
- Check image quality and relevance
- Identify duplicates and test data
- Mark assets with missing or incorrect annotations
- Note any privacy or compliance concerns
3. Choose your deletion method
- Individual deletion — Use for small numbers of specific assets
- Bulk deletion — Use for large groups of assets
4. Perform the deletion
Execute the deletion carefully:
- Double-check asset selections
- Review the deletion count before confirming
- Confirm you have a backup if needed
5. Verify the results
After deletion:
- Check the updated asset count
- Verify remaining assets are correct
- Review dataset insights for updated statistics
Best practices
Always download your dataset before performing large-scale deletions
Test deletion workflows with small selections before bulk operations
Double-check which assets are selected before confirming
Remember that deleting assets also removes all their annotations
Keep notes on why assets were removed for future reference
Identify assets to keep before starting deletion workflows
Impact of asset deletion
Understanding what happens when you delete assets:
Immediate effects
- Assets removed — Deleted from dataset immediately
- Annotations deleted — All associated annotations permanently removed
- Dataset count updated — Total asset count decreases
- Storage freed — Deleted assets no longer consume storage
Training and model impact
- Completed training — Models already trained keep their performance
- Training history — Past runs reference deleted assets but remain viewable
- Future training — Deleted assets won't be available for new training runs
- Active runs — In-progress training may be affected
No recovery option
- Permanent deletion — Cannot be undone through the interface
- No recycle bin — Deleted assets are immediately removed
- Restore from backup — Only option is re-uploading from saved backups
Common use cases
Quality control cleanup
Remove poor quality data to improve model training:
- Review assets for quality issues
- Identify blurry, corrupted, or off-topic images
- Delete individual assets using the action menu
- Verify dataset quality improved
Duplicate removal
Clean up duplicate or nearly-identical assets:
- Identify duplicate assets in your dataset
- Select duplicates using checkboxes
- Use bulk delete to remove all duplicates at once
- Confirm unique assets remain
Privacy compliance
Remove assets containing sensitive information:
- Identify assets with privacy concerns
- Use bulk selection if multiple assets affected
- Delete assets permanently
- Document compliance action
Dataset reorganization
Clean up datasets during project restructuring:
- Download full dataset as backup
- Identify assets to remove for reorganization
- Use bulk delete for efficiency
- Upload assets to correct datasets as needed
Troubleshooting
Cannot delete assets
- Check permissions — Ensure you have edit access to the dataset
- Browser issues — Try refreshing the page
- Active training — Some operations may be restricted during training
Deleted wrong assets
- No undo available — Deletion is permanent
- Restore from backup — Re-upload if you downloaded the dataset
- Re-upload originals — Use original files if available
- Re-annotate — Annotations must be recreated
Selection not working
- Click checkbox — Make sure to click the checkbox, not the asset thumbnail
- Refresh page — Clear any stuck states
- Browser compatibility — Ensure using a supported browser
Next steps
Learn how to remove individual assets
Delete multiple assets at once
Add new assets to your dataset
Backup your data before deletions
Analyze your dataset composition
Manage and review annotations
Need help?
We're here to support your VLMOps journey. Reach out through any of these channels:
Updated about 1 month ago