Turn Your WiFi Router into a Motion Capture Studio

Ever feel like your WiFi router is just sitting there, quietly shuffling bits around? What if you could repurpose it into something that feels like sci-fi—like a full-body motion capture system? That’s exactly what the wifi-densepose project does, and it’s a fascinating example of using everyday hardware in ways it was never intended.

Most motion capture systems require specialized cameras, suits with sensors, or expensive depth-sensing hardware. This project bypasses all of that by leveraging the radio signals that are already bouncing around your space. It’s a clever hack that turns a common piece of networking gear into a tool for 3D pose estimation.

What It Does

In simple terms, wifi-densepose uses the Channel State Information (CSI) from WiFi signals to estimate human poses in 3D. When WiFi radio waves travel through a space, they get distorted by the objects—and people—in their path. By analyzing these subtle distortions with a deep learning model, the system can infer the positions of a person’s limbs and torso, creating a skeletal motion capture output.

It’s essentially repurposing your router as a passive radar system for human movement. No cameras, no special wearables—just the existing wireless signals.

Why It’s Cool

The cleverness here is in the implementation. The project uses a model based on DensePose (a system for mapping all pixels on the surface of a human body), but instead of visual data, it’s trained on WiFi CSI data. This is a significant shift from traditional computer vision approaches.

Key highlights:

• Privacy-preserving: Unlike cameras, WiFi signals can’t capture identifiable facial features or fine-grained textures. It’s a more privacy-sensitive way to monitor movement.
• Hardware you already own: It works with common WiFi routers and network interface cards that support CSI extraction (like certain Intel WiFi cards).
• Works in low-light and through walls: Radio waves aren’t blocked by visual obstacles like walls or darkness, opening up unique use cases.
• It’s just neat: There’s something inherently exciting about using a communication protocol for spatial sensing. It feels like a hack in the best possible way.

How to Try It

Ready to experiment? The full code, pre-trained models, and setup instructions are available on GitHub.

1. Check the repo: Head over to github.com/ruvnet/wifi-densepose.
2. Review the requirements: You’ll need specific hardware (a router and a receiver NIC that can capture CSI data) and software dependencies (like PyTorch).
3. Follow the setup: The repository provides guidance on data collection, model training, and running inferences. It’s a research project, so be prepared for some hands-on configuration.

If you’re not ready to dive into the hardware setup, spend some time reading the paper and the code. The architecture and data processing pipeline are educational for anyone interested in wireless sensing or alternative inputs for deep learning.

Final Thoughts

This project is a brilliant reminder that the devices around us have untapped potential. While it’s still a research prototype and not a plug-and-play product, it points toward a future where environmental sensing is baked into our infrastructure. As a developer, it sparks ideas—could this be used for fall detection in healthcare, interactive installations, or non-intrusive activity monitoring in smart homes?

It’s the kind of project that makes you look at your boring old router and wonder, “What else are you capable of?”

Follow for more interesting projects: @githubprojects