Create a Dataset


User Guide

1 min


Annotab Team

Welcome to Annotab Studio!

In this tutorial, we'll guide you through the process of getting started with Annotab Studio. Don't worry, it’s incredibly straightforward.

Understanding Annotab Studio's Structure

In Annotab Studio, your work is organized into Datasets, each of which functions as an independent project. Users with the Member role can view and work within the Datasets they are assigned to. Meanwhile, users with the Admin role can view, edit, and manage all available Datasets.

Before you start

Your role will impact how you interact & use Dataset.

  • As Guest, you can only view invited Dataset and may not perform any other action.

  • As Member & Admin, you can create & manage a Dataset.

View more details about your role & permissions here.

Let's Get Started

  1. Creating a New Dataset

Begin by navigating to the Dataset module and clicking on the \"Create New Dataset\" button.


  1. Setting Up Your Dataset:

a. Title,Dataset Split and Instruction: Provide a Title and Instruction for your Dataset. This information will be visible to all members. Choose train/test/validation set split ratio during this step.


b. Defining Workflow: By default, your Workflow will include 3 following steps. In upcoming update, you will be able to create & customize your own workflow.


c. Uploading Data: Upload your data files. Currently supported file types include JPG, PNG, and WEBP.


d. Defining Classes: Define your classes by clicking \"Create new class\". Alternatively, you can skip this step and proceed to finish creating your project first.


  1. Completing Your Dataset:

Congratulations! You have now created your Dataset. Click on \"Next\" and your Dataset will be ready for use.


What's next

After creating your Dataset, you can proceed to annotate your data. Access your data by going to the \"Data\" section and clicking on the specific data you want to annotate.


View Also

  • Import & Manage Data

  • Annotate

  • Analytics

  • Generate & Export Data


Nov 8, 2023

Found helpful?