- Searching for similar data rows is now a one-click experience in Catalog. You can save similarity searches as a slice and combine similarity searches with other filters in Catalog. To learn more, visit our docs on similarity search.
- Labelbox now supports up to 5,000 vector annotations (i.e., point, polyline, bounding box, polygon) in the image editor. This improvement will help you label large-scale images with many objects of interest. This updated limit has been added to our Limits page.
- You can now view the global key on a data row and filter data rows by global key in the UI. To learn more see our docs on global keys.
- Free tier customers can now upgrade to the standard tier within the usage tab. Once you upgrade, you can manage your plan, view invoice history, update payment methods, and cancel your subscription via a billing portal.
- For all customers within the Free, EDU & Starter tiers, you now have the new Workflows feature enabled for all of your new projects. To learn more about how Workflows + Batches + Data row tab work together, visit our migration guide.
- You can now visualize your automation efficiency score in the Labelbox UI. This new metric is designed to help customers understand how well they are using model-assisted labeling to increase labeling efficiency. To learn more, visit our docs on Automation efficiency score.
- We’ve simplified the import format for uploading a JSON file through both the UI and the Python SDK (v3.30.0). With this standardized snake case format, you can seamlessly upload your data with a unified experience across both upload methods. You can see the new format in these docs: Image, Video, Text, Geospatial, Documents, Audio, DICOM.
The latest version of our Python SDK is v3.31.0. See our full changelog in Github for more details on what was added recently.
client.clear_global_keys()to remove global keys from their associated data rows
- Added a new attribute
ClassificationAnswerfor Model Error Analysis
project.create_batch()to work with both data_row_ids and data_row objects
- Added step to
project.create_batch()to wait for data rows to finish processing
project.setup_editor()multiple times no longer resets the ontology, and instead raises an error if the editor is already set up for the project
- create_data_rows, create_data_rows_sync, create_data_row, and update data rows all accept the new data row input format for row data
- create_data_row now accepts an attachment parameter to be consistent with create_data_rows
- Conversational text data rows will be uploaded to a json file automatically on the backend to reduce the amount of i/o required in the SDK.