> For the complete documentation index, see [llms.txt](https://docs.biodock.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.biodock.ai/start/path-2-train-a-fully-automated-ai-model/label-and-train.md).

# Label your first images

*What you'll get: your first tiles labeled and marked **Fully labeled**, ready to train on.*

{% hint style="info" %}
Labeling needs a desktop or laptop. AI labeling is up to 5× faster in Chrome or Edge than in Safari.
{% endhint %}

Training is now on its own page: [Train, check and improve](https://docs.biodock.ai/start/path-2-train-a-fully-automated-ai-model/train-check-and-improve.md). This page covers labeling.

## 1. Label your first tiles

<figure><img src="https://3806122466-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fk3DfNQKoMy1JXdaDLvpG%2Fuploads%2Fgit-blob-8deafcec185012952928c892aaab9d97e85c989c%2Fdocs-start-pen-fully-labeled.gif?alt=media" alt="The class Nucleus is picked, the Pen outlines a nucleus point by point, the other nuclei in the tile are outlined, then Fully labeled is ticked in the tile&#x27;s corner and the tile border turns green"><figcaption><p>Outline every object in a tile, then tick <strong>Fully labeled</strong> in its corner. The border turns green.</p></figcaption></figure>

1. Click your class in the left sidebar, or press its number (for example **1**). Three tools appear under it: **AI Detect**, **AI Select** and **Pen**.
2. Zoom in on one tile and outline every object in it, including objects cut off by the tile's edge.
3. Hover over the tile and tick **Fully labeled** in its corner. The tile's border turns green.
4. Do the same for more tiles, on more than one image. Use the image picker at the top of the sidebar, or **← Prev** and **Next →**, to change image.

Which tool to use:

* **AI Detect** (**D**) is the fastest. Drag a tight box around one typical object, and it outlines the look-alikes in the tile. Filter its guesses with the sliders, then click **Accept**. **Apply to all tiles** searches the whole image. It works for object classes only.
* **AI Select** (**S**) fills the gaps. Click a tile once to prepare it, click inside an object, and press **A** to keep the outline.
* **Pen** (**P**) is for anything the AI gets wrong. Click around the edge point by point, or drag along it.

How to use each tool, and how to select, edit, split and delete objects, is in [Labeling tools](https://docs.biodock.ai/projects-and-models/labeling-tools.md). Every shortcut is in [Labeling shortcuts](https://docs.biodock.ai/projects-and-models/labeling-shortcuts.md).

Hold **Space** and drag to move around. Hold **Shift** to hide every outline for a moment. Press **Esc** to put a tool down. There is no save button: everything saves as you go.

### The rules that matter

1. **Finish one tile at a time.** Label every object in it, including the ones cut off at the edge. It is normal for objects to be cut by tiles: when the model runs, Biodock joins them up again. Don't change the tile size to avoid it.
2. **Only tick Fully labeled when the tile is complete.** Only fully labeled tiles are used for training. Anything you didn't outline in a fully labeled tile teaches the model that it is background.
3. **Be precise.** Follow the real border of each object, and switch to the Pen when the AI is off. A few precise tiles teach more than many careless ones.

The **Tile Borders** legend at the bottom of the image explains the border colours. The image picker shows how many fully labeled tiles each image has.

## 2. Other things you can change while labeling

{% tabs %}
{% tab title="Classes" %}

#### Add a class

Click **Add new class** under your size group. If the new class is a very different size, or is a region rather than an object, add a new size group instead (see the **Size groups** tab). Only project owners can add classes.

#### Rename, recolour or delete a class

Click the gear on the class for **Rename class**, **Change color** or **Delete class**. You can also click the coloured circle to change the colour. A model version keeps the colours it was trained with. **Deleting a class deletes every object labeled with it.** Only project owners see this menu.
{% endtab %}

{% tab title="Size groups" %}
A size group is a set of classes that share one tile size and one segmentation type. Each size group trains its own model, and your model uses them together.

To add one, click **Add size group** at the bottom of the class list, and choose its segmentation type and tile size. Only project owners can add size groups.

Add a size group to:

* Find regions (such as tissue areas) and objects (such as cells) in one model.
* Find objects of very different sizes in one model, such as organoids and cells, or crypts and goblet cells.
* Label sparse classes (a rare cell type) apart from dense ones.

Each extra size group adds to the work of training and analysis. See [Configuring your model](https://docs.biodock.ai/projects-and-models/configuring-your-model.md) to nest one size group inside another.
{% endtab %}

{% tab title="Images" %}

#### Switch image

Click the current image at the top of the sidebar to see every image and how far along it is. You can filter and sort the list. Or use **← Prev** and **Next →**.

#### Add images

Leave the labeler with the **X** at the top left, open the **Data** tab, and click **Add from files**.
{% endtab %}

{% tab title="Channels" %}
Click **Channels** at the top of the sidebar to see one card per channel.

* **Solo** shows one channel on its own; **Show all** brings the others back.
* Untick a channel to hide it.
* Drag **Brightness** and **Contrast** when a channel is dim. **Discard** puts every channel back.

**With one size group, this only changes your view**, not your images and not what the model trains on. To change what the model sees, use **Settings** → **Appearance & channels** on the project page. See [Project settings](https://docs.biodock.ai/projects-and-models/project-settings.md).

**With more than one size group**, the panel also has **Save**, which saves your channel settings for training, and **Enable size group customization**, which lets each size group have its own settings. The panel then changes the size group you have selected. **Revert to project settings** removes every size group's own settings.
{% endtab %}
{% endtabs %}

## Next: train your model

When you have a few fully labeled tiles on more than one image, train your first model.

{% content-ref url="/pages/LptaKha8syHNhZKFfnV0" %}
[Train, check and improve](https://docs.biodock.ai/start/path-2-train-a-fully-automated-ai-model/train-check-and-improve.md)
{% endcontent-ref %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

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

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.biodock.ai/start/path-2-train-a-fully-automated-ai-model/label-and-train.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.
