Every WiFi router is already a sensor. It constantly measures how radio waves bounce, bend, and fade across a room just to keep your connection stable — and then throws that data away. RuView's bet is that this discarded Channel State Information encodes enough about the human body to recover much of what a camera would see, without the camera. That quietly reframes ambient WiFi from a networking detail into a privacy-preserving alternative to optical surveillance.
What Sets It Apart
- Commodity hardware, not lab gear. It runs on a mesh of four to six ESP32-S3 microcontrollers at roughly $9 each, rather than research-grade NICs. That moves WiFi sensing out of academic papers and into a weekend build costing tens of dollars.
- One signal, many readings. The same CSI stream drives 17-keypoint skeletal pose estimation, contactless breathing and heart-rate monitoring, occupancy counting, and fall detection — capabilities usually spread across separate dedicated sensors.
- Works where cameras can't. It sees through drywall and in total darkness, and stores no imagery at all, so the privacy and data-retention surface is fundamentally smaller than a camera or always-on microphone.
How the Signal Becomes Spatial
A human body subtly distorts the radio waves passing through and around it; models trained on those distortions reconstruct pose and vital signs from the resulting CSI patterns. The approach builds on Carnegie Mellon's WiFi DensePose research, packaged here as a Rust-heavy edge stack so inference can run on the microcontrollers themselves rather than a server.
Who It's For
Great fit if you want to prototype contactless health monitoring, elder-care fall detection, or camera-free occupancy sensing, and you're comfortable flashing ESP32 firmware and tuning models to your own space. Look elsewhere if you need certified medical-grade vitals, plug-and-play consumer hardware, or dependable accuracy out of the box — CSI sensing is highly environment-sensitive and needs per-room calibration, and the headline accuracy figures come from controlled datasets, not an arbitrary living room.
