Create an Annotation Task

To create a annotation task an annotation, click the Create Annotation button located in the top-right corner of the table section of your Annotation. This will open a form where you need to provide the required details.

Create Annotation

Create Annotation

Step 1: General Information

  • Annotation Task Name: Enter a unique and descriptive name for your annotation task to easily identify it later.

  • Dataset: Select the dataset you wish to annotate from the list of datasets associated with your project.

  • Dataset Version: Choose the specific version of the dataset that will be used for annotation.

  • Labels: Manually add label names or upload a .txt file containing predefined labels. These labels are essential for categorizing the data during annotation.

Create Annotation

Step 2: Guidelines

Upload a detailed guideline document for the annotators to follow during the annotation process. These guidelines are crucial for ensuring consistency and quality in the annotations across the team.

Create Annotation

Step 3: Labeling Configuration

Configure how the labeling will be carried out by choosing between manual human annotation and machine learning-assisted annotation:

  • Human Annotation: Assign specific users as labelers for manual annotation. You can specify multiple labelers and allocate a percentage of the dataset that each labeler is responsible for.

Create Annotation

  • ML-Assisted Annotation: If you prefer to use AutoML for annotation, select the compute resource to be used for the task. This method can significantly speed up the annotation process, especially for large datasets.

Create Annotation

Step 4: Review Setup

To ensure high-quality annotations, assign one or more reviewers for quality control:

  • Assign Reviewers: Choose reviewers who will be responsible for checking the annotations and approving them before they are finalized. Similar to labelers, you can assign multiple reviewers and set the percentage of annotations that each reviewer will evaluate.

Create Annotation

Once you’ve completed all the steps, click the Submit button to begin uploading your annotation. By following these steps, you can create a well-structured annotation task that leverages both human expertise and machine learning to achieve high-quality results. The platform’s flexible configuration options allow you to tailor the annotation process to meet your project’s specific needs.

For more details, refer to the Annotation Management Guide to explore advanced options and best practices for handling datasets on our platform.