February 1, 2022

Release notes

App

Added

Model Diagnostics now supports text classification and tiled imagery.
Model assisted labeling
- Now available to all paid Labelbox customers.
- Now supports Video Bounding Boxes
You can see your license tier from the Account tab.
The Document editor (beta) is available to all customers. You can use this new editor to create bounding boxes around key characters, concepts, or page sections on PDFs for Optical Character Recognition (OCR).
You can now view, create, and deactivate your webhooks directly in the Labelbox app. This means you no longer need to use GraphQL to set up your webhooks. Go to the Account settings page to try it out.
The Models dashboard allows you to visualize your Models and Model Runs in the Labelbox UI. You can also compare Model Runs against a set of Ground truth predictions.
The new DICOM editor is now available in closed beta. You can now label DICOM data in 3D space with segmentation masks If you’d like to sign up for early access, please sign up HERE.
Video segmentation is available in Open beta.

Fixed

Uploading instructions via the UI or the SDK (project.upsert_instructions(<path_to_file>)) should both now work as expected.
Search by Asset ID in Data Rows Tab works as expected.

Future

On the following dates, the Labels tab will be removed from the Labelbox UI. You will be able to find the same functionality in the Data Rows tab.
- February 15, 2022 for Free & EDU tiers.
- March 18th, 2022 for Starter, Pro, and Enterprise tiers.

Our latest version of the Python SDK is v3.12.0. See our full changelog in Github for more details on what was added recently.

Added

Fixed