Supervisely annotation format

Detailed explanation of how we store annotations in JSON format for images, videos, point clouds, DICOMs.

In Supervisely you can annotate data from several mediums: images, videos, 3D data. To accommodate and systemize annotations for all of these formats, we created our own JSON-based Supervisely Annotation Format, which we describe in detail in this section.

Table of contents

1. Project Structure

2. Project Classes and Tags

3. Objects

4. Tags

5. Image Annotation

6. Video Annotation

7. Point Cloud Annotation

8. Point Cloud Episode Annotation

9. Volume Annotation