Try for Free
PRICING cloudOn-prem deployment
Streamline Your Annotation Process with CVAT's Auto Annotation using Skeletons

Are you tired of spending countless hours annotating complex images with skeletons for tasks like sports analytics, animation, or healthcare? CVAT (Computer Vision Annotation Tool) has got you covered with its Auto Annotation with Skeletons feature that simplifies this meticulous task. In this guide, we'll walk you through how CVAT's Auto Annotation with Skeletons can make your annotation process simpler and faster.

The Power of Skeleton Annotation

Skeleton annotation is crucial in various fields where tracking body movement or posture is essential. Whether you're analyzing sports performance, creating lifelike animations, or monitoring patient movements in healthcare, skeleton annotation plays a pivotal role. However, manually annotating skeletons can be time-consuming and prone to errors. That's where CVAT's Auto Annotation feature comes into play.

Creating a Skeleton task

CVAT now supports the integration of models for automatic skeleton detection, including a newly added model capable of detecting 133 points as defined by the COCO Whole Body dataset. Let's see how it works step by step:

1. Task Creation: Start by creating an annotation task specifically for skeleton annotation.

2. Model Selection: During task creation, you will encounter the "Add from model" option. Use this option to select a model from CVAT's list of supported models and add labels directly from it.

3. Labels configuration Optional): You have the flexibility to configure the selected label according to your specific project requirements.

4. Image Upload: After adding labels, you can upload the images that you want to annotate.

Automatic Annotation with Skeletons

The process is pretty straightforward: open the job, go to the tools bar, and click on AI Tools. The model you need is in the Detectors section; select it from the dropdown list.

You will see the label mapping menu, where you can adjust labels either for the entire label or point by point. Here, you can also remove points if necessary or add them back. 

Once you're satisfied with the label mapping, click Annotate

CVAT will now use the selected model to automatically annotate skeletons in your images.

A Note on Visibility

It's important to note that if some points are not visible in a certain frame (e.g., they are off-screen), the automatic annotation may not detect them.

Improvements in Annotation Workflow

Save Time: With automatic annotation with Skeletons, you can instantly annotate multiple shapes (persons) in the image. No need to waste time manually labeling them one by one.

Accuracy: Automatic annotation with Skeletons is precise and accurate, but it doesn't limit you. After the initial annotation is done, you can easily make adjustments if needed.

Future Use: Once you've completed annotating your dataset, you can export the annotations quickly and accurately.


With CVAT Auto Annotation with Skeletons, you can significantly streamline the annotation process for complex images, saving time and ensuring accuracy, this feature will help you achieve precise results efficiently.

Happy Annotating!

Not a user?  Click through and sign up here

Do not want to miss updates and news? Have any questions? Join our community:






December 13, 2023
Go Back