> ## Documentation Index
> Fetch the complete documentation index at: https://avala.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Work Batches

> Manage annotation sessions, work units, and task assignment

This guide covers work batches in Avala -- how to group sequences into manageable units, assign them to annotators, and track progress.

## What Are Work Batches?

A **work batch** is a group of sequences assigned to an annotator as a single unit of work. Batches provide structure to large annotation projects by breaking them into manageable pieces.

Instead of assigning an entire dataset to one person, you create batches that:

* Distribute work evenly across your annotation team
* Set clear scope for each annotator's session
* Enable progress tracking at a granular level
* Allow parallel work on the same dataset

## Creating Work Batches

### From the Project View

1. Navigate to **Projects** → your project → **Batches**
2. Click **Create Batch**
3. Configure the batch:
   * **Name**: Descriptive label (e.g., "Highway Scenes Batch 3")
   * **Sequences**: Select which sequences to include
   * **Assignee**: Choose an annotator (or leave unassigned)
   * **Priority**: Normal or High
   * **Due date**: Optional deadline
4. Click **Create**

### Automatic Batch Creation

For large datasets, use automatic batching:

1. Go to **Batches** → **Auto-Create**
2. Set parameters:
   * **Batch size**: Number of sequences per batch (e.g., 50)
   * **Assignment mode**: Round-robin, random, or manual
   * **Annotators**: Select the team members to distribute to
3. Click **Generate Batches**
4. Review the generated batches and confirm

<Tip>
  Automatic batching distributes sequences evenly. If you have 500 sequences and set a batch size of 50, Avala creates 10 batches.
</Tip>

### From a Smart View

You can also create batches from filtered views:

1. Apply filters to your sequences (e.g., status, tags, metadata)
2. Select the filtered results
3. Click **Create Batch from Selection**

## Assigning Batches to Annotators

### Direct Assignment

1. Open a batch from the **Batches** tab
2. Click **Assign**
3. Select an annotator from the team member list
4. The annotator receives a notification and sees the batch in their work queue

### Reassignment

If an annotator is unavailable:

1. Open the batch
2. Click **Reassign**
3. Select a new annotator
4. Optionally add a note explaining the reassignment
5. The new annotator picks up where the previous one left off

### Self-Assignment

When enabled in project settings, annotators can claim unassigned batches:

1. Annotator goes to **My Work** → **Available Batches**
2. Reviews available batches
3. Clicks **Claim** on a batch to self-assign it

## Tracking Batch Progress

### Batch Dashboard

The batch dashboard shows an overview of all batches in a project:

| Column       | Description                      |
| ------------ | -------------------------------- |
| **Name**     | Batch name                       |
| **Assignee** | Annotator assigned to the batch  |
| **Status**   | Current batch status             |
| **Progress** | Sequences completed out of total |
| **Due date** | Deadline, if set                 |
| **Created**  | When the batch was created       |

### Progress Indicators

Each batch shows a progress bar:

* **Gray**: Sequences not yet started
* **Blue**: Sequences in progress
* **Yellow**: Sequences in review
* **Green**: Sequences completed

### Detailed Batch View

Click a batch to see:

* List of all sequences with individual statuses
* Time spent per sequence
* Issues flagged during annotation or review
* Activity timeline

## Session Management

### Starting a Session

When an annotator begins work:

1. Go to **My Work** → select a batch
2. Click **Start Session**
3. The first available sequence opens in the annotation viewer
4. A session timer begins tracking time

### Pausing and Resuming

* **Pause**: Click **Pause Session** in the toolbar. Progress is saved, and the annotator can return later.
* **Resume**: Go to **My Work** → select the batch → **Resume**. The viewer opens to where the annotator left off.

### Completing a Session

When the annotator finishes all sequences in a batch:

1. The last sequence is submitted
2. The batch status changes to **In Review** (if review is configured) or **Completed**
3. A summary shows:
   * Total time spent
   * Number of sequences completed
   * Number of annotations created

## Batch Statuses

| Status          | Description                                                              |
| --------------- | ------------------------------------------------------------------------ |
| **Unassigned**  | Batch is created but not assigned to anyone                              |
| **Assigned**    | Batch is assigned to an annotator but work has not started               |
| **In Progress** | Annotator is actively working on the batch                               |
| **Paused**      | Work is paused; can be resumed                                           |
| **In Review**   | All sequences are submitted and pending review                           |
| **Rework**      | Reviewer rejected one or more sequences; annotator needs to correct them |
| **Completed**   | All sequences are approved                                               |

```
Unassigned → Assigned → In Progress → In Review → Completed
                              ↕              ↓
                           Paused         Rework → In Progress
```

## Best Practices for Batch Sizing

| Dataset Size       | Recommended Batch Size | Reasoning                                   |
| ------------------ | ---------------------- | ------------------------------------------- |
| \< 100 sequences   | 20-30                  | Small enough for a single session           |
| 100-500 sequences  | 30-50                  | Balanced between overhead and manageability |
| 500-2000 sequences | 50-100                 | Larger batches reduce management overhead   |
| > 2000 sequences   | 100-200                | Large batches with periodic check-ins       |

### Sizing Considerations

* **Annotation complexity**: Complex annotations (dense segmentation, many objects per frame) warrant smaller batches
* **Annotator experience**: Newer annotators benefit from smaller batches with more frequent review
* **Deadline pressure**: Smaller batches complete faster and allow earlier review cycles
* **Data variety**: If sequences vary significantly, keep batches smaller so reviewers can calibrate per batch

<Warning>
  Avoid creating very large batches (500+ sequences). They make progress tracking difficult and delay the review cycle since the entire batch must be submitted before review begins.
</Warning>

## Next Steps

* Configure [Quality Control](/docs/annotation/guides/quality-control) for review workflows
* Understand [Sequence Statuses](/docs/annotation/reference/sequence-statuses) for tracking individual sequence states
* Set up [Team Permissions](/docs/annotation/guides/team-permissions) for annotator and reviewer roles