Jump toUpdate content

Smart Labeling Quickstart

Reviewed on 04 November 2021Published on 04 November 2021

Scaleway Smart Labeling is a platform that allows you to annotate images or videos, either by yourself or via a team of annotators. Smart Labeling currently uses Intel’s CVAT as its interface: open-source Computer Vision Annotation Tool. Smart Labeling facilitates the use of CVAT in a cloud-native manner, with labeling software running on-demand, annotations safely persisting in an automatically backed-up database, and datasets stored in cost-efficient S3 buckets.

Our solution comes with everything you need to start annotating. Only the designated Admin needs to create a Scaleway accout - annotators can simply log in directly to the publicly-accessible CVAT annotation interface without needing their own Scaleway accounts.

Requirements:
Note:

Currently, Smart Labeling only supports the annotation of image and video file formats. If you have other files stored in the linked buckets, you will not be able to view them in the annotation interface, but their presence will not impact the application.

How to create a Smart Labeling workspace

  1. Click Smart Labeling in the AI section of the side menu. The Smart Labeling splash screen displays:

  2. Click Create a labeling workspace. The workspace creation wizard displays.

  3. Enter the required information for your labeling workspace:

    • Enter a Name and an optional description for the workspace.
    • Select the Object Storage bucket(s) you want to use with your workspace.
    • Enter a username for the admin user.
    • Enter the email address of the admin user.
    • Enter a secret password for the admin user.
  4. Confirm by clicking Create a labeling workspace. Your workspace will be created.

Once ready, the workspace displays in the list of your Smart Labeling workspaces.

How to access your workspace

Click Smart Labeling in the AI section of the side menu. The list of your workspaces displays:

The information of your workspace displays:

Tip:

Note that what CVAT refers to as the superuser, who has full access to the administration panel of the annotation server, is the same person as the admin user of the labeling workspace. You can use these credentials to login.

How to access the annotation server

  1. From your Smart Labeling workspace, click Go to my annotation server in the workspace overview. The CVAT interface displays.

    Tip:

    Alternatively you can use the Annotation server URL to access the server. You can find it in the workspace information block. This URL will take you (or the annotator) to the CVAT annotation interface.

  2. Enter the credentials for the admin user of the Smart Labeling workspace and click Login. The CVAT dashboard displays:

  3. Create Tasks to annotate datasets of your choice with image-level tags, as well as points, bounding boxes, polygons, and 3D boxes and assign them to your users.

  4. Users can login into the CVAT interface to start annotating images in their tasks.

    Important:

    Annotators can create users and login to the annotation server directly from this interface.

How to assign tasks to annotators

Although each annotator can immediately use the account to log into the annotation server, they will not have access to the annotation tasks you have created, unless you assign the task to them, or you change their permissions by going into the administration panel of the CVAT server.

  1. Click your username in the upper right corner and click Admin page in the drop-down menu.

  2. Select the tasks and roles you want to add to each user and validate from the Django admin page.

    Tip:

    If you want to learn more about CVAT user groups and permissions, check out their documentation.

How to delete a Smart Labeling workspace

  1. Click Smart Labeling in the AI section of the side menu. The list of your workspaces displays.

  2. The information of your workspace displays.

  3. Scroll down to the Delete your labeling workspace section.

  4. Click Delete and type DELETE in the pop-up. Then click Delete this labeling workspace to permanently delete your workspace.

    Important:

    Deleting your labeling workspace deletes all your annotations. Make sure you have exported them, to avoid data loss.

    Note:

    Deleting the labeling workspace does not delete the associated S3 buckets. To avoid incurring Object Storage costs, it is recommended to delete the buckets separately if you no longer need the data.