Viewing the status of tasks (new)
Updated at April 26th, 2023
View task list
The Tasks page is used to search for specific tasks and view a range of information about those tasks.
By default, tasks are sorted by last submitted date and created date. The results are paginated. They contains the following information:
- ID: The task ID.
- Created: The date that the task was created.
- Last Submitted: The date that the task was last submitted by an annotator.
- State: The status of the task.
- Priority: The priority value that has been assigned to the task. The default is 0, but can be changed to any positive or negative number. When tasks are allocated to the next available agent, the task with the lowest priority number will be loaded first.
Filtering Tasks
Find a specific set of tasks by opening the Filter menu. Results update as filter values are changed. The results are paginated.
Viewing a task in read-only mode.
To view a task in the workspace in read-only mode, click on the row for the task you want to visualize. This lets you see how the task is progressing without any risk of accidentally changing anything. You can even look at a task while it is being worked on by an annotator — you'll see all their progress up to the most recent save.
Tip
If you know the project ID and task ID, you can jump right to this page by going to https://app.sama.com/projects/[projectID]/tasks/[taskID].
Comparing sets of answers in the read-only workspace
To quickly toggle between sets of annotations completed by the annotators or model predictions (also known as default answers uploaded during task creation), click on the "Submission history" dropdown menu in the navigation bar. Select the last option, Compare Answers. This will open a menu where you can select the answer to visualize. D stands for default answers, and the following options are manually annotated answers, according to the workflow step (the letter) and the round (the number). Rounds increment when there is a rejection and the task needs to be annotated again.
While toggling between answers, you will keep your context (frame, zooming & panning, selected shape) so that you can easily compare answers across submissions.
For Raster masks, viewing model predictions is not supported.