Skip to content

Teams & workspaces

Supervisely allows you to manage and develop datasets, models, custom extensions and many other entities. It is important to understand how we organize users, resources and access permissions.

Teams

Team is a group of users and resources.

Users in the same team share the same resources, like projects and models.

When you login to Supervisely, your last team will be selected. At every moment you work in a particular team and all new items like projects you create will be created in that team. You cannot create entities like projects outside of a team or have the same project in multiple teams at once (you can share them with Explore, Clone and Links, read below).

You can switch teams at any moment or create a new team in the left menu.

Members

A user in a particular team is called "member". Every user can be a member of one, many or even zero teams, but a team cannot exist without at least one member. Members can leave the team at any moment, but only team's admin can remove other members from the team.

Roles

Each team member has a role in the team that specifies what kind of actions they can do. For example, admin can invite new members, while developer cannot. The same user can have different roles in different teams.

At the moment the following roles are available:

  • Admin: Has full access in the team. Admin can invite new team members and remove entities created by the other team members.
  • Developer: Similar to the admin, but can only remove entities created by themself and cannot invite new members to the team.
  • Annotator: Has access only to a single page, Labeling Jobs.

Every team must have at least one admin, but can have two or more.

Cleanup

In order to free up space in the current group, you can go to the trash bin

All remote Projects, Models and Python Notebooks are located here.

To delete or restore an item you need to select it by clicking the check mark next to it. Then click "Delete forever" or "Restore"

Workspaces

When multiple users are collaborating together in the same team it is usually convenient to have a way to split results into separate "rooms" called "workspaces".

Workspace is filter inside a particular team that allows to quickly switch between different sets of experiments. There are three resources that are split between workspaces:

  • Projects and datasets
  • Neural networks
  • Tasks

A workspace is a part of a team. You cannot have a workspace outside of a team. Every team must have at least one workspace, but can have two or more.

At every moment you work in particular workspace and all projects, neural networks and tasks you create become a part of the current workspace and the current team.

Workspaces do not provide any access permissions or roles mechanisms, unlike teams.

Sharing

While having your resources organized in convenient, sometimes you need to share your resources between teams and users. There are three ways to do it.

Explore

You can share projects, neural networks and plugins with the whole instance using the Explore section. Choose "Publish" in the context menu to add items to make them accessible for everyone. Users in different teams see the same items on the Explore page. Other users cannot change items in Explore, but can clone them to their current team and workspace.

Clone

Clone can be used directly without Explore. Choose "Clone to..." in the context menu and select the team to which you want to clone the item. You can only select teams that have you as a member.

To clone an item directly to a team that doesn't have you as a member, create a unique link to share the item. When you open it in another team or workspace, you will see a conformation to clone a particular item.