Description of the "darwin dataset export" option from the CLI

Creates a Release for the given Dataset. By default, the release will only include the completed items from the dataset. Although this feature allows you to filter by specific class types, you need to keep in mind that you won't be able to create a Release with the same level of customization as you would with the UI.

darwin dataset export cars nice_cars --class-ids 1439 1438
Dataset cars successfully exported to andreas-team/cars:nice_cars

Positional arguments :

  • dataset: name of the dataset
  • name: name of the release to be created

Optional arguments:

  • --class-ids CLASS_IDS [CLASS_IDS ...]: List of annotation class ids. If present, it will only include items that have annotations with a class whose id matches.
  • --include-authorship: Each annotation contains annotator and reviewer authorship metadata.
  • --include-url-token: Each annotation file includes a url with an access token.Warning, anyone with the url can access the images, even without being a team member.



Anyone with the URL will be able to access the images, even without being a team member.



Currently there is no way for a user to get an annotation class's id, either via the UI or the CLI. To this effect, you can try to use our SDK, or ask support.
We are working to improve the user experience and we will move away from requesting object ids in CLI commands, so this is going to change in the future.

The above example creates a release called nice_cars with today's date, containing all completed DatasetItems that have annotations with class ids of 1439 or (inclusive OR) 1438.

Once the release is created, you can then download it via darwin dataset pull {DATASET}.

Next up

Having trouble understanding the docs? Have a refresh on V7 Concepts!