Google Open-Sources A New Hand Pose Detection Model

This new version can predict keypoints on both hands simultaneously.

Google shared a new version of their hand pose detection model which brings improved accuracy for 2D, novel support for 3D, and the new ability to predict keypoints on both hands simultaneously. "Support for multi-hand tracking was one of the most common requests from the developer community, and we're pleased to support it in this release," wrote the team in a blog post.

You can now try a live demo of the new model here. Google noted that this model is a new version of a previous model which allowed predicting 21 keypoints, but could only detect a single hand. The team also shared an article that discusses the new model and how you can get started. You can find the code here.

Don't forget to join our new Reddit page, our new Telegram channel, follow us on Instagram and Twitter, where we are sharing breakdowns, the latest news, awesome artworks, and more.

Published 21 November 2021

Arti Sergeev

Business Head