Body Detection & Pose Estimation Using WiFi

Researchers proposed a new model that can estimate the dense pose of multiple people using WiFi signals. It maps the phase and amplitude of WiFi signals to UV coordinates within 24 human regions.

To produce UV coordinates, the approach uses raw CSI signals cleaned by amplitude and phase sanitization, two-branch encoder-decoder network that performs domain translation from the CSI samples to 2D feature maps, and a modified DensePose-RCNN architecture that estimates the UV map.

While body detection is nothing new and is done by RGB cameras, LiDAR, and radars, it can be affected by occlusion and lighting or requires special equipment. This model uses a WiFi signal together with deep learning architectures and doesn't need much to work. 

The researchers believe that their model opens up new possibilities for low-cost, widely accessible, and privacy-respecting algorithms for human sensing.

Read more about it here and don't forget to join our 80 Level Talent platform, our Reddit page, and our Telegram channel, follow us on Instagram and Twitter, where we share breakdowns, the latest news, awesome artworks, and more.