Home

Welcome to wai-annotations

wai-annotations is a Python library for converting annotation datasets between various annotation formats of data domains such as images and audio. With its command-line tools, you can, e.g., create conversion pipelines or split dataset batches into subsets.

The library also offers plugins for reading/writing videos, preprocessing data, for generating statistics and visualizations, as well as incorporating predictions made via a Redis backend (e.g., using deep learning models).

• Installation - how to install the library
• Usage - running the library from the command-line
• Examples - common use-cases
• Docker - using wai.annotations via pre-built docker images
• Architecture overview - description of the internals
• Plugin Guide - adding new domains/formats/conversions
• Domains - supported domains
• Plugins - available plugins
• Glossary - explanations of terms

wai.annotations is being developed as part of the User-friendly Deep Learning project.