---
title: Labeling guide
tier: all
type: guide
order: 134
order_enterprise: 134
meta_title: Label and annotate data
meta_description: Label and annotate data to create bounding boxes, label text spans, set up relations. Filter and sort project data for machine learning dataset creation.
section: "Create & Manage Projects"
Label and annotate your data with the open source data labeling tool, Label Studio. After you set up your project and labeling interface and import your data, you can start labeling and annotating your data.
You can also collaborate with other annotators to improve the quality of your labeled data.
Some labeling tasks can be complicated to perform, for example, labeling that includes text, image, and audio data objects as part of one dataset and labeling task, or creating relations between annotations on a labeling task. This section includes guidance on how to perform more complex labeling tasks, such as labeling with relations, overlapping regions, selected tasks, or changing labels.
From a project, click Label All Tasks to start labeling all tasks. To label the tasks as they are filtered and sorted in the data manager, select Label Tasks As Displayed instead.
You can also label a specific task in the Quick View or Preview by clicking it from the project data manager view, but you won't automatically see the next task in the labeling queue after submitting your annotations.
You can also select the checkboxes next to specific tasks and then click Label $n Tasks to label the selected number of tasks. For example, select the checkboxes for 5 different tasks, then click Label 5 Tasks to label only those 5 tasks.
!!! note
When labeling tasks, you should not open the label stream (meaning to click Label All Tasks) simultaneously in two tabs. This could result in you receiving the same task twice, which can circumvent project settings that address annotator overlap.
Annotate a section of the data by adding a region.
When you label with bounding boxes and other image segmentation tasks, or when you're highlighting text for NLP and NER labeling, you might want to label overlapping regions. To do this easily, hide labeled regions after you annotate them.
You can select multiple regions while labeling to make changes to them together.
ctrl and click each region that you want to select. You can select regions on the object that you're labeling or in the Regions sidebar. Select a range of regions in the Regions sidebar by clicking the first region in the list that you want to select and holding Shift while you click the last region in the list that you want to select.u to deselect it.You can duplicate a region to create many identically-sized polygons, rectangles, or ellipses.
ctrl + d to duplicate the region, or the sequence of ctrl + c and ctrl + v to copy and paste the region. Duplicated regions appear in the exact location as the region being duplicated. If you're using a Mac keyboard, use cmd instead of ctrl.You can change the label of an existing region.
You can delete labeled regions, such as bounding boxes or text spans, if needed.
After labeling a task, you can delete the annotation by clicking the trash can button to delete an annotation on the task. If you haven't saved your annotation yet, click the X icon to reset the task and remove your labeling activity.
You can also delete all annotations on a task from the project page. See Delete tasks or annotations.
You can create relations between two results with both directions and labels. To add labels to directions, you must set up a labeling config with the relations tag. See more about relations with labels in the Tags documentation.

Figure 1: Add relations between annotations
After you relate two annotation regions, you can modify the relation in the Relations section of the Results sidebar.
- To change the direction of the relation, click the direction button between the two related regions.
- To add labels to the direction arrow indicating the relation between two annotations, click the vertical ellipsis button next to the two related regions to add your predefined labels. You must have a label configuration that includes relations to do this.
When annotators skip a task, the task no longer appears in the labeling queue for that annotator. Other annotators still see the task in their labeling queue.
When annotators are not finished with their annotation work but would like to exit their labeling flow to, for example, pause the lead time calculation on their annotations, they can click the drop-down menu next to Submit and then select Submit and Exit (or Update and Exit). This will submit the current annotation and redirect them to the Projects page.
If they are not done with their annotation, then they can simply check to make sure their draft was saved in the history panel and then navigate to the Projects page to stop the lead time calculation from running.

In both Label Studio and Label Studio Enterprise, you can label tasks with collaborators. Tasks are locked while someone performs annotations so that you don't accidentally overwrite the annotations of another annotator. After the other annotator finishes with the task, it can appear in your queue for labeling if the minimum annotations per task is set to more than one. By default, tasks only need to be annotated by one annotator.
If you want to label tasks more than once, even if the minimum annotations required is set to one, do the following:
To label tasks multiple times while the minimum annotations required is set to one, do the following:
1. In the data manager for the project, click a task to open the quick labeling view.
2. Click the + icon next to the task annotation ID to open an annotation tab.
3. Label the task.
4. Click Submit to save your annotation.
5. Click the next task in the data manager to open the quick labeling view for that task and repeat steps 2-4.
If you have a machine learning backend set up to get interactive preannotations, you can choose whether to use those predictions while you label.
Click the settings icon when labeling to configure the labeling interface to suit your labeling use case.
For example, keep a label selected after creating a region, display labels on bounding boxes, polygons and other regions while labeling, and show line numbers for text labeling.

Figure 2: Customize the labeling interface
You can also modify the layout of the screen, hide or show predictions, annotations, or the results panel, and hide or show various controls and buttons.
If you want to perform advanced image labeling, follow these examples and guidance for assistance.
You can add multiple types of regions to image annotations. You can add any of the following:
- Rectangles
- Ellipses
- Keypoints
- Polygons
- Brush masks
To add different types of regions to your image annotations, follow this example.
Create a custom template for your labeling interface using the following example:
<View>
<Image name="image" value="$image" />
<Rectangle name="rect" toName="image" />
<Ellipse name="ellipse" toName="image" />
<KeyPoint name="kp" toName="image" />
<Polygon name="polygon" toName="image" />
<Brush name="brush" toName="image" />
<Choices name="choices" toName="image">
<Choice value="yes"></Choice>
<Choice value="no"></Choice>
</Choices>
<Labels name="labels" toName="image" fillOpacity="0.5" strokeWidth="5">
<Label value="building" background="green"></Label>
<Label value="vehicle" background="blue"></Label>
</Labels>
</View>
This example makes rectangles, ellipses, polygons, keypoints, and brush masks available to the annotator, along with image classification choices of yes and no, and region labels of building and vehicle.
You can add a rectangle or an ellipse to your image with just two clicks, or double click to create a polygon, rectangle, or ellipse.
If you accidentally select a point on an image while creating a polygon, just double click to remove the erroneous point and continue creating the region. There must be at least three points on the polygon to be able to remove a point.
When you're annotating images, you can create regions without applying labels.
This can be helpful for two-step labeling, where you want one annotator to create regions and another annotator to label the regions.
If you make a mistake when labeling with the brush mask, you can erase it. You must select a brush region in the sidebar before you can erase any part of it.
If you want to completely remove a region and start over, delete the region instead of erasing it. Erasing a region does not delete it.
The Outliner is a replacement for the existing interface with a multiple-panel approach. This feature improves the user experience to fit more data in these two columns (**Outliner** and Details) and provides more options when working with regions. For example, Image segmentation is the only type of labeling that supports custom controls on the details panel.
Outliner is the area where you can see all the details about annotation, regions, and labeling history. It is split into two panels, named Outliner and Details. Figure 3 shows the appearance of the Outliner panel on the right of the image and Details panel on the left of the image.

Figure 3: View of Outliner and Details panels view
Panel actions
There are two panels which is the outliner for the regions list. These panels can be collapsed, moved, swapped to the sides, dragged, dropped, and expanded to provide a full-screen view. For example, if you navigate to label stream, you will have more space in this mode.
The Outliner and Details panels are collapsible and detachable, so you can arrange them the way you want. Figure 4 shows the Collapse details button which allows you to collapse the Details panel and use the remaining portion of the screen.

Figure 4: Collapsed details button
Now, the Details panel is collapsed, and you can expand it by clicking the Expand Details icon.
Figure 5: Details panel in collapsed view
The details panel is displayed in expanded view to explore the feature.

Figure 6: Details panel in expanded view
To collapse the Outliner panel, click on the Collapse outliner icon.

Figure 7: Collapse outliner button
Now, the Outliner panel is collapsed, and you can expand it by clicking the Expand Details icon.
Figure 8: Outliner panel in collapsed view
The Expand Details icon expands the Outliner panel for further actions.

Figure 9: Outliner panel with expand outliner icon
The Outliner panel is displayed in expanded view to explore the feature.
Figure 10: Outliner panel in expanded view
These panels are detachable, so you can place them wherever you want on the screen, and you can resize them.

Figure 11: Details panel in floating dock view
Region editor
You can edit regions with a focus on labeling. You can zoom in and precise control over the numbers and dimensions of each region attribute. The image region details like height, rotation, and so on can be changed using the inputs in the Details panel. Now, you can collapse all panels and get the maximum working area. This feature design makes some actions clear in the product user interface.

Figure 12: Edit regions using different attributes
!!! note
In the upcoming releases, video segmentation will be available to control keyframes, animations, and custom metadata.
The Comments box is available inside the outliner.

Figure 13: Comments box
!!! note
Use Outliner to work with larger annotation tasks (many bounding boxes in one image, larger videos, and so on).
To add a region in the OCR transcription templates, draw a rectangle, and you can see a text box that appears in the Outliner panel to enter OCR text. The new functionality allows you to select a region and the Details panel changes. You can experience the following:
Regions grouping or sorting
You can group or sort regions using the following functionality from the drop-down list.
!!! warning
Group by Label and Group by Tool options does not work for multiple image scenario or multiple labeled type scenario; only the Group Manually option is available for this mode.
You can also order your regions by Order by Score or Order by Time options.
![]()
Figure 14: Group by label using sorted by score

Figure 15: Group by tool using sorted by score

Figure 16: Group by label using sorted by score

Figure 17: Manual grouping
Prediction indication of the region
This capability shows whether an ML model created the region or not.
Relations list
One region belongs to another one. It is typically done for NLP type of text. For example, consider the city name as region 1 and district name as region 2. Now, in the relations list hierarchy you will place the region 2 under region 1. Relation lists can be unidirectional or bidirectional between one region and another. You can snap one region into another region.
!!! note
The relations list does not support multi-level listing for regions.
Draw bounding boxes over another
!!! attention "important"
When creating bounding box annotations for images, you must not start a bounding box inside another bounding box because it would just drag the first box. Instead, it would create a new bounding box in front of the image when the rectangle tool is selected.
To create bounding boxes on top you can use two ways:
!!! note
This option is not always visible.
Now, you should be able to drag when the rectangle tool is not selected. Remember that when the rectangle tool is selected, you can make rectangles even if it is on top of another rectangle.
OR
{% insertmd includes/annotation_ids.md %}