October 19, 2021

Release notes



  • Now you can use the Python SDK to import ground truth annotations directly to your Data Rows in Labelbox. This differs from MAL in that you don’t have to hit “submit” in the Editor for the annotations to be applied to your Data Row. This is helpful for getting started with Model Diagnostics or for migrating annotations from an external project.
  • Our Metadata feature for custom and reserved fields is now in GA.
  • We added a new schema tab that allows you to create custom metadata fields. To learn more about custom metadata and its value for helping with data selection check out the docs on Metadata & Catalog.
  • We improved the Catalog filters so that now you can define and import custom metadata to your DataRows and then find/select DataRows using filters.
  • We added an Enterprise Dashboard that provides a global view of all of your labeling operations across your Labelbox projects. From here, you can filter and schedule regular reports for your org (contact [email protected] to request access).

Python SDK

On 10/11/2021, we released a new version 3.7.0. It includes the following updates. Visit our changelog for details on earlier versions.


  • Search for data row ids from external ids without specifying a dataset
  • client.get_data_row_ids_for_external_ids()
  • Support for numeric metadata type


The following DataRowMetadataOntology fields were renamed:

  • all_fields -> fields
  • all_fields_id_index -> fields_by_id
  • reserved_id_index -> reserved_by_id
  • reserved_name_index -> reserved_by_name
  • custom_id_index -> custom_by_id
  • custom_name_index -> custom_by_name