Each unique project has a specific need when it comes to annotation tools. One might need either an image annotation tool, a text annotation tool, a video annotation tool or some combination of the above. With so many tools to choose from, finding the right fit can be a frustrating process. Not to worry though- we’ve rounded up a master list of the best open source annotation tools in 2022. Keep reading to find the best annotation tool for your unique needs!

Best Image and Video Annotators

CVAT

The Computer Vision Annotation Tool is a powerful and efficient image and video annotator. It is open source and web based, and though its user interface is not very intuitive, amateurs and professionals will be able to take advantage of it after getting over the learning curvet.

Key Features:

• Web based
• Maintained by Intel
• Annotation is semi-automatic

Pros

• Web based 
• Third Party Integrations
• Easy to deploy on local networks
• Tutorials available online
• Supports a variety of file types

Cons

• No customer support
• CVAT has to be maintained as it scales
• No pdf file support
• Learning curve; the UI may take several days to master

‍

Source code- https://github.com/opencv/cvat

‍

Labelimg 

Labelimg has been around for over 5 years, and is one of the popular, dependable tools for graphic image labelling on the web. It has a simple interface which is also pretty intuitive, making it pretty easy to work with.

Key Features

• Written in Python
• Graphical Interface is in QT
• Annotations are saved and exported as XML files
• Needs to be installed locally

Pros

• Can be used offline for added security
• Simple, intuitive interface, even for beginners

Cons

• Default version offers only one annotation type (bounding box)
• Needs to be installed locally

Does not offer video annotation

Source code- https://github.com/tzutalin/labelImg

‍

LabelMe

LabelMe is an open source dataset of digital images with annotations. Free to use, it was created by the MIT Computer Science and Artificial Intelligence Laboratory in 2008, and users are allowed to contribute to the library. It has a voluminous library, described by some as canonical.

Key Features

• Made by MIT Computer Science and Artificial Intelligence Laboratory

Pros

• Six different types of annotations offered
• Customizable UI
• Can be used both online and offline

Cons

• Files can only be exported and saved in JSON format
• No program management capabilities
• Low level of precision

Source code- https://github.com/tzutalin/labelImg

‍

OpenLabeling 

OpenLabeling is a sturdy tool for both image and video annotation in computer vision applications Created by João Cartucho, this tool was licensed in 2018.

Key Features

• Runs in Python
• Features a pre-trained model 

Pros

• Multiple annotation formats are available, for example PascalVOC and YOLODarknet
• Deep Learning feature available

Cons

• Both Python and OpenCV have to be downloaded to use this tool

Source code- https://github.com/Cartucho/OpenLabeling

Best Text Annotation Tools

YEDDA

Developed to annotate chunks of text, YEDDA is able to work in many languages including English and Chinese. Text, symbols and even emojis can be accurately annotated by this super tool.

Yedda also supports shortcut annotation which increases efficiency in annotating text by hand.

Key Features

• Runs in Python, requires Python preinstallation
• Supports shortcut annotation

Pros

• Collaborative capability
• User Interfaces both for administrators and annotators

Cons

• Users complain of a few bugs while using this tool
• Not available offline

Source code https://github.com/jiesutd/YEDDA

‍

ML-Annotate

Another popular open source text annotator, ML-Annotate is one of the first choices for many when it comes to text annotation. Developed by 

Key Features

• Runs on Python

Pros

• Administrative users can be added
• UI is completely customizable, with instructions for making modifications provided
• Supports multi-class,multi-label and binary labelling
• Can be used offline

Cons

• Data library not included.

Source code https://github.com/falcony-io/ml-annotate

‍

We hope this was helpful! If you’re still undecided, you can check out our table below for the summarised version of all the info above.

Image and video annotation comparative table

‍