In Model & Catalog gallery view, you can now see Entities/NER (named-entity recognition) annotations and predictions on text asset thumbnails.
You can now retrieve a slice’s data rows and all associated information programmatically via our Python SDK. From there, you can inspect the Data Rows and a) curate a batch using Data Row IDs in a slice, or b) add them to a Model Run as a training dataset. See our guide on Slices for more details.
We now provide error messages to indicate processing errors upon Data Row upload. These messages are intended to help you troubleshoot Data Rows that were not created properly. See Processing issues for a complete list of errors.
When you are configuring an ontology, you now have the option to reorder options in a nested ontology. Visit our docs on creating/modifying ontologies to learn more.
Our GraphQL API explorer has an updated UX, an option for dark mode, and offers a history log.
In Catalog, you can save filters as a Slice. The Slices feature enables you to easily access previously created filters so you only need to create the filter once. See our docs on Slices to learn more.
Our documentation has been restructured to make the data import formats for each annotation/asset combination easier to find and share (e.g., Image annotations).
The latest version of our Python SDK is v3.29.0. See our full changelog in Github for more details on what was added recently.
- Added new base
client.get_catalog_slice(id)to fetch a CatalogSlice by ID
slice.get_data_row_ids()to fetch data row ids of the slice
- Add deprecation warning for queue_mode == QueueMode.Dataset when creating a new project.
- Add deprecation warning for LPOs.
- Default behavior for metrics to not include subclasses in the calculation.
- Polygon extraction from masks creating invalid polygons. This would cause issues in the coco converter.
- Added warning for upcoming change in default project queue_mode setting
- Added notebook example for importing Conversational Text annotations using Model-Assisted Labeling
- Updated QueueMode enum to support new value for QueueMode.Batch =
- Task.failed_data_rows is now a property
- Fixed Task.wait_till_done() showing warning message for every completed task, instead of only warning when task has errors
- Fixed error on dataset creation step in examples/annotation_import/video.ipynb notebook
- Added deprecation warning for missing
- Updated docs for deprecated methods
- Use the
Projectto get and set the queue mode instead
- For more information, visit https://docs.labelbox.com/reference/migrating-to-workflows#upcoming-changes
- Use the
project.export_labelsto support filtering by start/end time formats "YYYY-MM-DD" and "YYYY-MM-DD hh:mm:ss"