Data annotation

On the SuperAnnotate platform, you can use various project types to annotate your items efficiently and produce the most high-quality training datasets. The project types you have at your disposal are LLMs and GenAI, Image, Video or Audio, Text, and Tiled Imagery.

Project types

Here are all of the project types that you can make use of on the platform. To get started on LLMs and GenAI Projects, you can read here.

Image

The Image project has a comprehensive suite of AI-assisted annotation tools to help make annotation more efficient. Automatically segment instances with the click of a button using Magic Select (SAM) and Magic Polygon, transcribe text from images with ease using Magic Box (OCR), and use Keypoint Workflows for skeleton annotation.

To set up an Image project:

1. In Projects, click on + New Project.
2. Select Image.
3. Enter a project name.
4. If applicable for your team, you can enable Annotate Similar to make use of the feature in the Explore tab (optional).
5. Click Create.
6. Begin importing your dataset into the project.

Video or Audio

You can upload a video or audio file with the Video or Audio project. For video projects with Frame Mode on, you can easily track and segment moving items in videos using Autotrack.

To set up a Video or Audio project:

1. In Projects, click on + New Project.
2. Select Video or Audio.
3. Enter a project name.
4. Video Projects only - If applicable for your team, you can enable Frame Mode to view your videos in frames instead of a timeline.
5. Click Create.
6. Begin importing your dataset into the project.

Text

In the Text project, you can classify different text snippets in your documents and connect them together for context using Relationships.

To set up a Text project:

1. In Projects, click on + New Project.
2. Select Text.
3. Enter a project name.
4. Click Create.
5. Begin importing your dataset into the project.

Tiled Imagery

In the Tiled Imagery project, you can upload higher-resolution images for more accurate, in-depth annotation.

To set up a Tiled Imagery project:

1. In Projects, click on + New Project.
2. Select Tiled Imagery.
3. Enter a project name.
4. Click Create.
5. Begin importing your dataset into the project.

Classes

You can use classes and their attributes to define the instances you create in your projects.

The attributes of a class are separated into attribute groups, and they all help you specifically define and differentiate between instances for an accurate annotation process.