JAKARTA - Using only a WiFi router, researchers at Carnegie Mellon University, United States (US) developed a method for detecting three-dimensional shapes and human body movements in a room.

The researchers used an Artificial Intelligence (AI) model called DensePose to estimate body position from 2D photos. DensePose was developed by London-based researchers and Facebook AI researchers.

Armed with a pair of entry-level routers that cost around $30 each. The routers are installed at opposite ends of the room, and the computers measure the signal transmission at 2.4GHz between the access points.

Then, they applied the model to a feature map created from recorded Wi-Fi signals. After training, AI can convert that data into a 3D mesh representation. When superimposed over a photo, the approximate body position can be very accurate.

According to research recently published on arXiv servers pre-press under the title DensePose From WiFi, using a Wi-Fi antenna can be a lot cheaper than a regular RGB camera or LiDAR if tracking people somewhere.

"The study results reveal that our model can predict solid poses of multiple subjects, with performance comparable to an image-based approach, by utilizing the Wi-Fi signal as the only input," said researchers Jiaqi Geng, Dong Huang, and Fernando De la Torre in their paper.

"This paved the way for low-cost, widely accessible, and privacy-guarding algorithms for human sensing."

Furthermore, the researchers also think their Wi-Fi approach to human imaging can be applied to home healthcare, where patients such as the elderly or those with disabilities may not want to be monitored with cameras in places like bathrooms or with sensors and other tracker devices.

Most importantly, the Wi-Fi monitoring system is not affected by dim light or obstructions, such as walls. Plus, it only requires very inexpensive technology, using equipment most people already have at home.

"In fact, most households in developed countries already have Wi-Fi at home, and this technology could be scaled up to monitor the well-being of older people or simply identify suspicious behavior at home," the researchers said.

In addition, researchers also show this technology can actually be a boon for privacy. The camera can capture identifiable images, but the Wi-Fi signal only produces a human-shaped web.

This could lead to a more privacy-conscious way to monitor a room for someone trespassing, where they won't know who's there, but can tell where they were and what they were doing.


The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)