Introducing Masks

V7’s pixel perfect semantic segmentation tool means you can create the annotations that make the most sense for your use case.

When to Use Masks

Masks and Polygons are both used for object segmentation, and both classes have their own strengths & use cases. The choice between Masks and Polygons depends on the specific requirements of the task and the characteristics of the objects being annotated.

📘

A file can contain both raster and polygon annotations.

Use Cases

Semantic masks can be more suitable than polygons for some use cases. These use cases could be:

  1. To catch fine details - masks allow you to annotate at pixel-level accuracy
  2. For complex objects - masks allow you to annotate items with complex boundaries with more flexibility
  3. When a pixel should only have one identity - masks do not allow pixels to have multiple identities like polygons

Mask Set Up

In the Classes menu, you can create a Master annotation type by selecting Mask in the dropdown menu!

Creating Masks

The best way to create a raster annotation is with the Brush tool, in combination with the Zoom tool.

We recommend you set the timeline bar to 1

Don't forget to use the shortcut keys [ and ] to increase and decrease the size of the brush, and e and f to swap between eraser and brush!

📘

Mask annotations are available for all data types supported by V7.

Masks are displayed in the annotation panel underneath annotations.

👍

Use SAM to Save Time

You can now use SAM to create Mask annotations on images & videos. Please see this page to see how to do this: Using SAM in V7

Limitations

Undo/Redo

In this initial release of Masks, we currently do not have Undo/Redo. Please use the Brush Tool to adjust your Mask annotations.

Exporting Masks

Masks can only be exported in Darwin JSON V2 format.

Please refer to the Darwin JSON schema to understand how Masks are represented in exports.