BlazePose: On-device Real-time Body Pose tracking

By Valentin Bazarevsky, Ivan Grishchenko, Karthik Raveendran, Tyler Lixuan Zhu, Fan Zhang, Matthias Grundmann. arXiv.org, 2020

861 citations recorded by the source corpus.

We present BlazePose, a lightweight convolutional neural network architecture for human pose estimation that is tailored for real-time inference on mobile devices. During inference, the network produces 33 body keypoints for a single person and runs at over 30 frames per second on a Pixel 2 phone. This makes it particularly suited to real-time use cases like fitness tracking and sign language recognition. Our main contributions include a novel body pose tracking solution and a lightweight body pose estimation neural network that uses both heatmaps and regression to keypoint coordinates.

Subfields: Computer Vision and Pattern Recognition, Human-Computer Interaction

View Paper: https://www.semanticscholar.org/paper/c9bcea08fb81c041ed6d2b7576d8f0e47c1c850f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Tao Jiang
2. Ning Ma
3. Kai Chen
4. Rui Han
5. Li Zhang
6. Chengqi Lyu
7. Peng Lu
8. Yining Li

Citations in the corpus

• RTMPose: Real-Time Multi-Person Pose Estimation based on MMPose (2023, methodology): “MediaPipe [3] BlazePose-Lite BlazePose N/A 256× 256 N/A N/A 29.”