WiFi Signals That See You — Without a Camera

RuView turns ordinary router signals into a real-time human pose detector, and it's moving fast. Forty-nine thousand stars in a repository that most people hadn't heard of six months ago. That number alone should make you pause — but the reason behind it is more interesting than the count. Setting RuView comes from ruvnet, the same builder behind several agentic-AI experiments that have quietly accumulated serious GitHub attention over the past year. The timing here is not accidental. Two currents are colliding at once: the cost of ESP32 microcontrollers (small, programmable chips the size of a postage stamp) has dropped below five dollars, and computer-vision researchers have been refining a technique called DensePose — originally developed to map human body positions from photographs — for nearly a decade. RuView is the project that asks: what happens if you run DensePose not on a camera feed, but on WiFi radio signals bouncing off a person's body? The answer, apparently, is something that 49,630 developers decided was worth starring. The Story Here is the concrete scenario. You place a cheap ESP32 board — the kind sold in bulk for IoT hobby projects — somewhere in a room. It broadcasts and listens to WiFi signals. When a person moves through that space, their body subtly distorts the radio waves, the same way a hand passing through a flashlight beam changes the light pattern on the wall. RuView processes those distortions in real time, written in Rust (a systems programming language chosen for its speed and memory safety, both critical when you are running inference on a microcontroller), and reconstructs a skeletal pose of the person: where their arms are, whether they are sitting or standing, even basic vital signs like breathing rate. No camera. No microphone. No pixels stored anywhere. A practical example: imagine a care facility for elderly residents where staff need to detect falls or irregular breathing during the night, but residents understandably refuse video surveillance in their rooms. A single RuView-equipped ESP32 board on the ceiling handles presence detection and vital-sign monitoring silently, privately, without recording a single image. The system flags anomalies to a monitoring dashboard; the resident never needs to wear a sensor. You can see a live demonstration of the interface at Cognitum.One/RuView, and the source is fully open at ruvnet/RuView on GitHub. The topic tags on the repo — agentic-ai, densepose, esp32, firmware, mcu, mincut, pose-estimation — tell the architecture story in shorthand. MinCut is a graph algorithm used to sharpen the signal separation; agentic-ai hints that the monitoring layer can make autonomous decisions rather than just logging data. This is not a weekend script. It is a small, dense system with real engineering ambition. The Insight The momentum here is not just about one clever repo. It is about a specific window that is opening right now: commodity hardware has finally become capable enough, and open-source AI research has finally become accessible enough, that a single developer can ship something that would have required a university lab and a government grant five years ago. RuView is a signal — pun intended — that the next wave of ambient intelligence will not come from companies deploying expensive LiDAR rigs or camera arrays. It will come from people flashing firmware onto five-dollar chips and pointing them at empty rooms. The project is four days from its last push as of this writing, actively maintained, and the star velocity suggests it is nowhere near peak discovery. The readers who go deep on this now — understanding the Rust firmware layer, the DensePose adaptation, the MinCut signal processing — will have a real head start when this pattern becomes a product category. If you have been watching the agentic-AI and edge-computing spaces converge, RuView is one of the clearest demonstrations yet of where that convergence lands in the physical world. Worth an afternoon. Possibly worth much more than that. More repos moving this fast, across every category, are tracked weekly at teum.io/stories. 한국어 요약 RuView는 5달러짜리 ESP32 칩 하나로 WiFi 신호를 분석해 카메라 없이 사람의 자세와 생체 신호를 실시간으로 감지하는 오픈소스 프로젝트입니다. Rust로 작성된 펌웨어, DensePose 기반 포즈 추정, MinCut 신호 처리가 결합된 구조로, 단순한 주말 해킹 프로젝트가 아닙니다. 현재 GitHub 스타 49,630개를 넘어섰고 아직 성장 중입니다. 노인 돌봄, 비디오 없는 보안, 스마트홈 등 응용 범위가 넓어 지금 파악해두면 유리합니다. No camera. No microphone. No pixels stored anywhere — just radio waves and a five-dollar chip doing things that used to require a university lab.