WiFi-3D-Fusion, an open-source system that uses wireless LAN signals to estimate human location and pose

The technique of detecting target behavior using wireless LAN signals is called CSI sensing (Wi-Fi sensing) , and in recent years, various researchers have published results in this field. Recently, an open-source system called ' WiFi-3D-Fusion ' has been released on GitHub, which can run CSI sensing locally and visualize the results in 3D animation.

GitHub - MaliosDark/wifi-3d-fusion: WiFi-3D-Fusion is an open-source research project that leverages WiFi CSI signals and deep learning to estimate 3D human pose, fusing wireless sensing with computer vision techniques for next-generation spatial awareness.
https://github.com/MaliosDark/wifi-3d-fusion

Research into sensing human posture and movement using Wi-Fi radio waves has been ongoing for some time, and many methods for creating 3D models of objects have been reported.

Introducing 'WiPose,' a new technology that uses Wi-Fi radio waves to create a 3D model of the posture and movement of a person behind a wall - GIGAZINE

It is possible to generate 3D images of humans using a Wi-Fi router in a normal home - GIGAZINE

'WiFi-3D-Fusion' is a project based on the following research published in 2024, which aims to perform CSI sensing locally and visualize it in 3D animation.

aiotgroup.github.io/Person-in-WiFi-3D/
https://aiotgroup.github.io/Person-in-WiFi-3D/

'WiFi-3D-Fusion' has a web interface like the one below, and it will create a 3D model of the detected object.

The developers claim that the web interface of 'WiFi-3D-Fusion' includes features such as 'manual camera control with mouse rotation and zoom,' 'ground noise visualization,' '3D skeleton rendering of detected people,' and 'live data streaming at 10 FPS.'

The developers of WiFi-3D-Fusion explained their motivation for developing the system, saying, 'The world is filled with invisible signals, and a huge amount of information flows through us every second, but most people are unaware of it.' They explained that this system could be useful in places where cameras fail.

Opinions about 'WiFi-3D-Fusion' have been exchanged on the social news site Hacker News, but some users have pointed out that 'the entire README has been heavily edited by AI (the final output is all AI), and worst of all, the images appear to be generated ( probably by GPT-4o)', raising doubts about the project's authenticity.

WiFi-3D-Fusion – Real-time 3D motion sensing with Wi-Fi | Hacker News
https://news.ycombinator.com/item?id=45020457

Others commented that the interface and code resembled Claude, the AI from Anthropic, and that there were some suspicious aspects to the README.