Labelbox documentation

January 17, 2020

Fixed

  • The warning message no longer appears when you modify your ontology mid-project in the new Image Editor. However, the warning may still appear if you have nested classifications.

  • When you remove all features on a label and hit save, the label will be marked as skipped.

  • When you update a previously skipped label with features or classification content, the label is no longer marked as skipped.

  • If a user adds or updates a feature on a label and does not explicitly hit “Save” when finished, it will no longer cause missing objects in the export.

  • You can now exit the Tasks dropdown by clicking on any other part of the page.

Added

  • You can now filter by Label ID, External ID, and Asset ID in the dropdown menu in the Activity section of the Labels tab within a project.

  • You can now turn on the feature class name overlay by selecting “Overlay object titles” under the settings icon in the image editor.

  • It’s now possible to re-assign a new class to a feature by clicking on a feature and selecting a new class from the left pane.

  • The general release of the Python SDK is now available. And we have open-sourced it so that you can extend it further or customize it to your needs.

Changed

  • When a user rejects a review of a label, that label will be entered into the review queue for re-review.

  • When you schematize your customization options, the API will raise a 400 error if there are over 1000 schema nodes.

Future

  • We will be releasing a beta of the video labeling editor in the coming months. Keep an eye out for future announcements!