ESP32 + WiFi + AI = See Through Walls for Under $25
ESP32 + WiFi + AI = See Through Walls for Under $25No cameras. No sensors. Just your WiFi router and some $8 hardware. It maps your body, tracks your movement, and reads your heart rate — through walls.
Someone just open-sourced software that sees you through walls using only WiFi signals. It’s called WiFi-DensePose (RuView). It maps your exact body pose in real-time. No cameras. No sensors. Just your living room router. 100% open source.
This started as a Carnegie Mellon University research paper that proved WiFi signals can reconstruct human body poses. RuView took that concept and turned it into an actual working system anyone can run. 31,700+ GitHub stars. MIT licensed. And it works on $8 hardware.
🧠 What This Actually Does — In Plain English
Your WiFi router sends signals around your house all the time. When those signals hit your body, they bounce back slightly differently — the way your arms are positioned, how fast you’re breathing, whether you’re standing or lying down — all of this changes the WiFi signal in tiny, measurable ways.
RuView reads those tiny changes and uses AI to turn them into a full body map. Think of it like sonar or echolocation, but using WiFi waves instead of sound.
What it can detect:
| Capability | What It Means |
|---|---|
| Body pose estimation | Tracks 17 body points (joints, head, arms, legs) in real-time — like a motion capture suit, but invisible |
| Vital sign monitoring | Reads your heart rate and breathing rate from WiFi signal disturbances — no wearable needed |
| Presence detection | Knows if someone is in a room, even through walls — no motion sensor, no camera |
| Through-wall sensing | All of the above works through walls. The WiFi signal doesn’t care about walls |
| Multi-person tracking | Can track multiple people at the same time |
| Fall detection | Detects if someone falls down — useful for elderly care |
| Sleep monitoring | Tracks breathing patterns while someone sleeps |
| Gesture control | Recognizes hand gestures for controlling devices |
All of this happens without a single camera or wearable device. Just WiFi.
📚 The Science Behind It — Carnegie Mellon University
This isn’t science fiction. It’s published research.
In January 2023, researchers at Carnegie Mellon University published a paper called “DensePose From WiFi” (arXiv:2301.00250). They proved that a deep neural network can map WiFi signal phase and amplitude to 24 human body regions — reconstructing full 3D body pose from WiFi alone.
Their setup used just 3 standard WiFi transmitters and receivers. The AI system identifies body joints, head position, torso, arms, and legs by analyzing how WiFi signals bounce off people. It works through walls, in darkness, and with multiple people in the room.
RuView took this research and built a practical, deployable system on top of it — written in Rust, running on cheap hardware, with a real-time dashboard and Docker support.
Original paper: arxiv.org/abs/2301.00250
⚙️ How to Try It — Three Options
Option 1: Simulation Mode (No hardware needed — just Docker)
docker pull ruvnet/wifi-densepose:latest
docker run -p 3000:3000 ruvnet/wifi-densepose:latest
Open http://localhost:3000 — you’ll see the full dashboard running with simulated data. This exercises the entire pipeline without any special hardware.
Option 2: Basic WiFi Sensing (Your laptop’s built-in WiFi)
Works on Windows with standard WiFi. Detects presence, motion, and coarse breathing rate. No pose estimation (that requires CSI hardware).
docker run --network host ruvnet/wifi-densepose:latest --source wifi --tick-ms 500
Option 3: Full Pose Estimation (ESP32-S3 hardware — ~$8 per node)
For the real deal — body tracking, vital signs, through-wall sensing — you need ESP32-S3 chips that expose Channel State Information (CSI). These cost about $8 each. Deploy 3-6 nodes for best accuracy.
docker run -p 3000:3000 -p 5005:5005/udp ruvnet/wifi-densepose:latest --source esp32
GitHub repo: github.com/ruvnet/RuView
Live demo: ruvnet.github.io/RuView
💰 Why This Matters — Real Use Cases
| Use Case | How It Works | Who Benefits |
|---|---|---|
| Elderly care / fall detection | Place ESP32 nodes in a parent’s home. Get alerts if they fall or stop moving. Monitor breathing at night — all without cameras | Families with aging parents |
| Smart home automation | Know which rooms are occupied without motion sensors. Trigger lights, heating, music based on actual human presence | DIY smart home builders |
| Home security | Detect intruders through walls. Know exactly where someone is in your house. No cameras to hack, no blind spots | Privacy-conscious security setups |
| Sleep tracking without wearables | Monitor breathing rate and body position while sleeping. No Apple Watch, no Fitbit, no wrist discomfort | Anyone who hates sleeping with a wearable |
| Gaming / gesture control | Control games or devices with hand gestures — no controller needed. Like a Kinect, but invisible | Gamers, AR/VR experimenters |
| Fitness / pose tracking | Track exercise form in real-time without a camera pointed at you. Great for home workouts | Home gym users |
| Disaster response | Detect people trapped under rubble using WiFi signals. RuView has a built-in disaster response module | Search and rescue teams |
| Retail / foot traffic | Track how people move through a store without creepy cameras. Heatmaps from WiFi signals | Store owners, event organizers |
| Research / academic | Build on top of the CMU paper. The entire system is MIT licensed and the training pipeline is open | Students, researchers |
🔧 Hardware You Need — Cheap and Simple
| What | Cost | Purpose |
|---|---|---|
| ESP32-S3 development board | ~$8 each | The sensor that reads WiFi CSI data |
| 3-6 nodes recommended | ~$24-48 total | More nodes = better accuracy and through-wall performance |
| Any computer with Docker | You already have this | Runs the processing server and dashboard |
| Standard WiFi router | You already have this | Provides the WiFi signal to analyze |
Important distinction: Your regular laptop WiFi only exposes RSSI (one number per access point) — good enough for presence detection and basic motion. Full body pose estimation requires CSI (Channel State Information) — that’s what the ESP32-S3 nodes provide.
The entire software stack is free and open source (MIT license). The only cost is the hardware.
📊 Project Stats
| Metric | Value |
|---|---|
| GitHub stars | 31,700+ |
| Forks | 4,200+ |
| License | MIT (use it for anything) |
| Language | Rust (fast, memory-safe) |
| Active since | March 2025 |
| Based on | CMU “DensePose From WiFi” paper (arXiv:2301.00250) |
| Hardware cost | ~$8 per node (ESP32-S3) |
| Docker image | ruvnet/wifi-densepose:latest |
Quick Hits
| Want | Do |
|---|---|
 Try it now (no hardware) |
→ docker run -p 3000:3000 ruvnet/wifi-densepose:latest → open localhost:3000 |
 Basic presence detection |
→ Use your laptop’s WiFi with --source wifi flag |
 Full body tracking |
→ Buy 3x ESP32-S3 (~$24 total) → flash firmware → deploy mesh |
 Read the research |
→ arxiv.org/abs/2301.00250 |
 Explore the code |
→ github.com/ruvnet/RuView |
Your WiFi router has been seeing through walls this whole time. Now you can too.
!