SuperAnnotate Docs

Annotation

Annotate

Use the following tools to annotate your pixel projects.

Brush

Select the brush tool, adjust the brush size, and outline your object.

Polygon

Select the polygon tool, and outline your objects using clicks at each corner of your shape.

Box Tool

Select the box tool and outline your object.

Rotated Box

Create a rotated box by drawing a line to set its width, then move the rotated box upwards or downwards to determine its size. To rotate the box, drag the rotate tool in a circular motion.

Ellipse

Select the ellipse tool, draw a line to set the diameter, and move the ellipse upwards or downwards to determine its size.

📘

  • A zoom level is maintained the same while switching between images.

  • Note that if you zoom in on an image, switch to another one where the zoom coordinates are missing, the zoom level will be reset to the default level.

📘

To move around an image while annotating, you can click and hold the scroll wheel and move the mouse, or you can use the W, A, S, and D keys. Press W and S to go up and down, and press A and D to go left and right.

Semantic and instance segmentation modes

You can set the segmentation mode in the Pixel editor. Choose between semantic segmentation and instance segmentation.

  • Semantic segmentation

    • The Pixel editor is in semantic segmentation mode by default meaning that all the instances that are assigned to the same class count as one. Start by creating an instance and assign it a class. Next, create other instances and assign them the same class. The instances you create count as one instance.
  • Instance segmentation

    • In the left panel, turn the Semantic/instance segmentation mode on by sliding the toggle to the right. In this case, the instances won’t be grouped together and each instance will count as one.

Smart segmentation

In the Images tab, select one or multiple images, select Smart segmentation, and choose a model: Generic or Autonomous driving. When Smart segmentation is done, go to the editor.

In the left panel, choose how much you want to segment your image under Number of segments. The higher the number you choose, the more your image will be segmented, and vice versa. Next, click on the segment you want to annotate, press the spacebar, and assign it a class.

Smart prediction

In the Images tab, select Smart prediction next to the image you want to annotate. Next, choose a model:

  • Instance segmentation (trained on COCO): Predicts your instance’s class label, bounding box, and binary mask. Use this model to detect countable objects, such as cats.
  • Panoptic segmentation (trained on COCO): Assigns all the pixels to classes. Use this model to detect countable objects, such as cars, and uncountable objects, such as the sky.
  • Semantic segmentation (trained on Cityscapes): Assigns all the pixels to classes. Use this model to annotate street scenes.

Go to the editor, click on the segment you want to annotate, and assign it a class.

Overwrite mode

Activate the Overwrite mode to add instances over annotated areas. The system will delete the annotated underlayers automatically.

Apply pre-annotations

To apply pre-annotations to an image, select Apply preannotation in the top panel then choose Yes. Note that the uploaded annotations will overwrite your current annotations. Learn how to upload pre-annotations here.

📘

Only a Team owner, Team admin, and Project admin can apply pre-annotations.

Adjust annotation

Use the Eraser Brush and the Eraser Polygon tools to fix your annotations.

Eraser Brush

Press Shift+X, and brush over the areas you want to delete.

Eraser Polygon

Press Shift+Z, and draw a polygon over the areas you want to delete.

Eraser ellipse

Press Shift+J, and draw a polygon over the areas you want to delete.

Eraser box

Press Shift+M, and draw a polygon over the areas you want to delete.

📘

Note that SuperAnnotate automatically saves your progress.

Approve/Disapprove instance

In the right panel, you can approve and disapprove instances of images with the following statuses: Completed, Quality check, and Returned.

To approve or disapprove an instance, select the corresponding thumbs up button or thumbs down button in the right panel.

📘

Approving an instance does not change the status of a Completed image.

If at least one instance is disapproved, the image will be automatically disapproved. In this case, images that have a Completed status will be sent for a quality check. The status of a Quality check or Returned image will remain the same.

📘

Only Team Owners, Team admins, Project admins, and QAs can approve and disapprove instances.

📘

QA automation

Use an error detection algorithm to detect annotation mistakes and import the information to SuperAnnotate via Python SDK for a faster quality assurance process.

Delete annotation

Select the annotation you want to delete, find it in the list in the right panel, and select X.
To delete all the annotations, select Revert (trash icon) in the top panel or press Ctrl+Shift+R.

📘

Select the keyboard icon in the bottom left corner or press Ctrl+K to get access to keyboard shortcuts.

Class

Assign class

You can assign a class to your annotation in five ways:

Method 1: Right-click on your annotation and choose a class from the drop-down menu. Note that the instances in the right panel will expand. To collapse the instances, select the down arrow next to the corresponding class.

Method 2: Select your annotation and choose a class from the drop-down menu in the right panel. Note that the instances in the right panel will collapse. To expand the instances, select the down arrow next to the corresponding class.

Method 3: Select Class selection and choose a class, create an annotation, and press the spacebar.

Method 4: Select the Eyedropper tool, click an object to sample its class. The Eyedropper tool will change into the Bucket tool. Next, select your unassigned object(s). The Eyedropper tool eases the class assignment of multiple objects.

Method 5: After applying Smart Segmentation to an image, choose a number of segments from the left panel. The higher the number you choose, the more your image will be segmented, and vice versa. Now select a segment, click the plus button in the right panel, choose a class.

Alternatively, you can assign a class to your object before annotating it. Follow these steps:

  • Select the class selection tool from the left panel.
  • Choose a class.
  • Annotate your object using any tool you want.

Your object will be immediately assigned to the class you’ve chosen.

Change class

You can change your annotation’s class in two ways:

Method 1: Right-click on your object and choose a class from the drop-down menu.

Method 2: Select your object, find it in the list in the right panel, select the ellipsis button, and move your object to a different existing class.

Objects

Show or hide objects

Select the eye button in the top panel or press R to show or hide the annotated objects.

Image

Filter images

Click the filter button in the image panel to filter the images by:

  • Assignee: No assignee, No assignee: QA, and No assignee: Annotator.
  • Annotation status: Any, Not started, In progress, Quality check, and Returned.
  • Prediction: Any, Not started, In progress, Completed, and Failed.
  • Segmentation: Any, Not started, In progress, Completed, and Failed.
  • Pinned images: To do that, select the pin button.

Select Reset > Apply to remove the filter.

📘

Press the right arrow to move to the next image and the left arrow to move to the previous image.

Add tags

Follow these steps to add tags to images:

  1. Select Tags in the left panel.
  2. Enter a tag in the text box.
  3. Click the + button or press Enter.
  4. Select Save.

Adjust settings

Adjust the settings of your image by dragging the slider to the left or to the right:

  • Image (transparency)
  • Fill (instance transparency)
  • Lines (segment transparency)
  • Brightness
  • Contrast

Approve/Disapprove image

In the top right corner, you can approve (thumbs up button) and disapprove (thumbs down button) images with the following statuses: Completed, Quality check, and Returned.

When you disapprove an image that has a Completed status, it will be sent for a quality check. However, the status of a Quality check or Returned image remains the same.

When you approve an image, its status remains the same.

📘

Only Team owners, Teams admins, Project admins, and Customers can approve and disapprove images.

Set image as complete

To set your image as complete, select Completeat the bottom of the right panel. Learn about project workflow.

Send image to QA

To send your image to the QA(s), select the down arrow at the bottom of the right panel and click Quality check. Learn about project workflow.

Request images

Annotators and QAs can request images by selecting the plus button in the Images panel. They will be automatically assigned up to 50 images and cannot have more than 50 images at a time. Annotators and QAs have to finish working on the assigned images to be able to request more images.

See image information

In the image panel, hover over the info (i) icon to see the image name and the names of the QAs and Annotators. You can copy and paste the image name. You can also hover over the icon on the left to see the image status.

📘

Only the Admin can see the names of the users that are assigned to work on the image.

Comments

Add comment

  1. Select Comment in the left panel or press C
  2. Click where you want to add the comment.
  3. Enter a text.
  4. Press Enter or select the send button.

A red dot will appear on the comment’s icon.

📘

Adding a comment does not change an image’s status.

Resolve comment

Select the comment and click the green arrow. Only Team owners, Team admins, and Project admins can resolve comments. The red dot on the comment icon will disappear.

Delete comment

Select the comment > Delete, and confirm your action. Only Team owners, Team admins, Project admins, and QAs can delete comments. The red dot on the comment icon will disappear.

Updated 6 days ago


Pixel


Suggested Edits are limited on API Reference Pages

You can only suggest edits to Markdown body content, but not to the API spec.