Open Source
Data Labeling Tool

  • — Simplicity built-in, no overcomplicated UIs
  • — Supports different datatypes
  • — Visually configurable from top to the bottom

Introducing version 0.7.0 - Cloud Storage Enablement

Load assets from AWS S3 or GCP. Read more in our blog.

Last version is 0.7.4.post0

10000+ Users

Data Scientists, Machine Learning Engineers, and aspiring developers are using Label Studio to produce innovative machine learning models for real-world applications.

Image Bounding Box

Audio Transcription

Dialog Analysis

Audio Regions

Named Entity Recognition

Quickstart

# Install the package
pip install label-studio

# Create a new project
label-studio init my_project

# Launch it!
label-studio start my_project

Check out the templates configured for specific data labeling needs.

Templates

Why Label Studio?

Configurable

Using HTML-like tags, you can quickly configure the UI for your particular needs. It's not just an image or text. It can be a pairwise comparison, multi-type classification, and beyond. Label Studio is a swiss army knife of data labeling and annotation.

Learn More

Multiple data types

Label Studio is type agnostic. The overall goal is to be able to process all data types. Right now it supports Audio, Text, Images, and HTML.

Each interface can include a mix of the types with different workflows.

Embeddable and Extendable

Label Studio includes a powerful embedding feature that enables you to embed entire UI in your pipeline. You can extend core functionality with new visual constructs and customize it for different use cases.

Read More

Compatible with your favorite framework

Looking to scale?

  • — Do you have more data than you can possibly label by hand?
  • — Do you want to distribute the labeling to multiple annotators?
  • — Do you want to control the quality of the annotations?
  • — Do you want to include your entire team into the process to achieve greater results?

We provide commercial offering built on top of Label Studio.

Quickstart