Tutorial 10: Export to LabelStudio

This tutorial shows how you can easily label all images of a tag from Lightly using the open-source data labeling tool LabelStudio.

What you will learn


You have a dataset in the Lightly Platform and optionally already chosen a subset of it and created a tag for it. Now you want to label all images of this tag using LabelStudio.

If you have not created your own dataset yet, you can use any dataset (e.g. the playground dataset) or follow the docs to create one.

Launch LabelStudio

Follow the documentation to install and start Labelstudio. Then create a new project and click on import. Now you should be in the import screen.

Import dialog of LabelStudio.

Export from Lightly in the LabelStudio format

  • Now open your dataset in the Lightly Platform.

  • Select the tag you want to export at the top. By default, the initial tag is already chosen.

  • Navigate to the Download page to see the different download options.

  • Within Export Reduced Dataset, select LabelStudio Tasks from the dropdown of the list of supported export formats. Specify an expiration duration giving you enough time to label all images. The tasks include a url pointing to the real images, thus allowing everyone with the link to access the images. This is needed for LabelStudio to access the images without needing to login.

  • After clicking the button ‘Export to LabelStudio Tasks’, they are downloaded as a single json file to your PC.

Import the tasks into LabelStudio

Now head back to LabelStudio and import the file you just downloaded. Either per drag-n-drop or browse your local files. Then finish the import.

Imported file into LabelStudio.

Start labeling

Now you can start labeling your images! To see them, you might need to change the type of the image column to ‘img’.

LabelStudio tasks fully imported and showing images.

Total running time of the script: ( 0 minutes 0.000 seconds)

Gallery generated by Sphinx-Gallery