# AI-assisted pre-labeling

### Description:

You can access the AI-assistant via a tab in your [project](/documentation/project/what-is-a-project.md) detail page. The AI-assistant allows you to automatically add labels to the images in your project based on the **convolutional neural network (CNN)** provided by **DATA**GYM. We offer you 80+ classes (ex., car, person, traffic light, etc.) to detect objects in your images. With a few click you can define which classes you need for your project and the respective labels will be added to your images. Thereby, you can **pre-label** the images to reduce the workload for your team and speed up the overall labeling process. [Here](/documentation/ai-assistant/object-classes.md) you can see the full list of classes supported by our CNN.

{% hint style="info" %}
The AI-Assistant is only available for users with a **Team Pro** subscription. Upgrade your [pricing plan](https://www.datagym.ai/pricing/) to make labeling images even faster! 
{% endhint %}

![](/files/-M9SVmHvk26FISpKlkif)

### Configure a label mapping:

Use the label-mapping table to define which objects you want to pre-label in your project. Therefore, you select a class from our CNN in the first column and a geometry from your [label configuration](/documentation/label-configuration/what-is-a-label-configuration.md) in the second column. Thereby, you can can assign the classes to the objects you want to find in your project. Please note that only [bounding box geometries](/documentation/label-configuration/entry-types.md#available-geometries) can be used for the pre-labeling. 

{% hint style="info" %}
You need at least one bounding box in you label config to configure the pre-labeling functionality.
{% endhint %}

![](/files/-M9SW_1FjK3kIIQ7AACT)

You can assign multiple classes to the same label configuration entry. In the example above, the classes `car` and `truck` are both assigned to the `Vehicle` geometry. Thereby, all cars and trucks will be labeled as vehicles in your project.

### Manage the pre-labeling process:

The buttons on top of the page allow you to **start** and **stop** the pre-labeling process. The pre-labeling only considers images in [tasks](/documentation/tasks/what-is-a-task.md) that are in the **WAITING** state. Make sure to move all task from **BACKLOG** to **WAITING** before you start the pre-labeling process. 

![](/files/-M9SWopocj-rJ50GTi_g)

After starting the process, a progress bar will keep you updated about the number of images that have already been labeled. Use the stop button to cancel the pre-labeling process. You can stop the pre-labeling process at any time and update the label mapping as you like before you continue.

![](/files/-M9SX00gl9l-rXS2rOJ3)

After the pre-label process finished, you can see the total number of images/tasks in your project that were labeled.

{% hint style="info" %}
Please note that tasks in your project are pre-labeled only once. You have to add or reconnect a specific dataset to start a new pre-labeling for the corresponding tasks.
{% endhint %}

### Pre-labeled images:

If an image was pre-labeled it is marked with a light-blue clipboard icon as shown in the image.

![](/files/-MC7NN38yqu7T1k8cwYf)

### AI-assistance limit:

Pre-labeling counts towards the AI-assistance limit of your pricing plan. For every pre-labeled image your remaining AI-assistance limit for this month will be reduced by one. If you try to label more images than you have Ai-assistance remaining you will be notified and can stop the process to reconsider the images you want to pre-label.

![](/files/-MC6yYwMbwZ_fQ6EVuw3)

If your remaining AI-assistance reaches zero, the start button is disabled and you won't be able to further pre-label images.

![](/files/-MC6ypAiE09EXTwFYmmc)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.datagym.ai/documentation/ai-assistant/ai-assisted-pre-labeling.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.