Updates
Migration guide: Export V1 to Export V2
General information about the Export V1 deprecation and the advantages of moving to Export V2.
Starting in April 2024, we will begin sunsetting Export V1 for UI and SDK for customers on a rolling basis. We encourage you to start transitioning your export workflows to using
To see a more detailed comparison of Export V1 and Export V2, visit this Google Colab notebook. To learn more about the advantages of Export V2, see Export overview.
export(), an improved method and format for exporting your data from Labelbox. You need to update your internal workflows to use the new export() method. Please look out for email notifications regarding sunset dates for your tier.
Export V2 offers the following advantages:
- More granular control over your data exports across the Labelbox platform and SDK, more data fields to export, and better robustness and performance.
- Ability to export specific data rows using the filters in Catalog and Annotate. For example, you can select only the data rows that have received new labels, metadata, or issues updates within the last 24 hours.
Deprecated SDK methods
The following Export V1 methods are deprecated and will be sunsetted in 2024. Labelbox will notify you of your sunset date based on your tier (see dates above).project.export_labels()project.label_generator()project.export_queued_data_rows()project.video_label_generator()dataset.export_data_rows()batch.export_data_rows()model_run.export_labels()
Fields and path mapping
To view a mapping of the Export v1 fields/paths to the equivalent fields/paths inexport(), view this sheet.
Comparison
| Export V1 | Export V2 | |
|---|---|---|
| Structure | Label-centric | Data row-centric |
| Export source | Annotate only | Catalog, Model, Annotate |
| Filters | None | Model run filters, Catalog filters, Data row tab filters, last_activity_at, label_created_at, data_row_ids |
| Customizable export payload | Not supported | Supported via filters |
| Naming convention | Pascal case | Snake case |
How to use export()
To learn how to export your data rows from a project via the UI, go to Export labels from project guide. To learn how to export your data rows via the Python SDK, go to our Export overview developer guide.Sample projects
| Media type | Project export | Model run export |
|---|---|---|
| Image | See sample | See sample |
| Video | See sample | See sample |
| Text | See sample | See sample |
| Geospatial | See sample | See sample |
| Documents | See sample | See sample |
| Conversational text | See sample | See sample |
| Audio | See sample | Not supported |
| HTML | See sample | See sample |
| LLM human preference | See sample | Not supported |
FAQ
What does this mean for projects that were created before export() was released?
What does this mean for projects that were created before export() was released?
This deprecation of exports v1 will not impact your existing projects.
What will happen on the sunset date? Will I be able to use Labelbox that day?
What will happen on the sunset date? Will I be able to use Labelbox that day?
Yes. Labelbox will disable Export V1 on the specified dates for each tier. You can still use Labelbox on the dates we disable Export V1 for your tier.