Hybrid & On-prem
Cloud data overview
Restrict data access by IP range
How to generate signed URLs
How to generate non-expiring signed URLs
Data import overview
Upload data via app
Import URLs via JSON
Asset metadata & image overlay (beta)
Model-assisted labeling (import annotations)
Real-time human-in-the-loop labeling
Custom label interface
Creating your first project
Project setup script
Python SDK FAQ
Intro to the GraphQL API
Data types overview
Bulk import requests
Legacy vs new editor ontology
Legacy vs new editor JSON exports
Legacy vs new editor hotkeys
Model predictions (legacy)
October 9, 2020
September 25, 2020
August 21, 2020
August 6, 2020
July 6, 2020
June 22, 2020
June 2, 2020
May 19, 2020
April 14, 2020
April 1, 2020
March 3, 2020
February 18, 2020
February 5, 2020
January 17, 2020
How it works
The Members feature allows you to invite individual users and set the permission settings for each member. If multiple users are collaborating on a project, Labelbox will distribute the data to the members with the appropriate access.
Members will only get unique data, meaning Labelbox will not allow multiple users to label the same data unless Auto Consensus is turned on.
Benefits of adding members
- Complete your projects faster. Labelbox automatically divides the data attached to a project to all the labelers which can help complete a project in a fraction of the time compared to only having one labeler.
- Manage performance. Track the total number of labels, time per label, and total time spent labeling for each member. This will help you manage the efficiency, cost, and complexity of your labeling project.
- Benefit from labeling consensus. With Auto Consensus turned on, a portion of your dataset will be labeled by multiple labelers. Labelbox will then compare the agreement for each of these labels across the multiple labelers and calculates a consensus score. Auto Consensus works in real time so you can take immediate and corrective actions towards improving your training data and model performance.
You may set permission levels for each user in one of two ways: at the organization level or at the project-based level. Organization roles and project-based roles are designed to be mutually exclusive. Here is a breakdown of the differences between the two:
A role assigned at this level will apply to all projects within your organization.
Users cannot be restricted from accessing certain projects at from the organization level.
If set, the user's project-based roles cannot be modified.
Roles at this level are assigned on a project-by-project basis.
Contains an additional "No access" permission setting to restrict a user from selected projects.
If set, the user's organization role cannot be modified.
With the exception of the "No access" permission which is only available as a project-based role, the role titles at the organization and project levels are the same: Labeler, Reviewer, Team manager, and Admin.
- A Labeler can label queued data and view their own work from the Activity table. A labeler cannot view labeler performance or view/modify any aspect of the project configuration or data sources.
- A Reviewer extends the Labeler role by being able to see and review all labeled data in the project.
- A Team manager extends the Reviewer role by being able to add and remove members inside of a project. The Team Manager can also view labeling performance for all project members.
- The Admin role can add and make modifications to projects, data, and labelers. Only organization admin users can export labels. By default, the user that signs up for the account has the role of Admin at the organization-level. Project-based admins cannot upload data.
See docs for adding/removing a member to learn how to assign organization and project-based roles to individual users.