stackpicks.dev
All posts
RuView — See Through Walls With WiFi and an $8 ESP32 in 2026
Emerging Tech·8 min read

RuView — See Through Walls With WiFi and an $8 ESP32 in 2026

RuView turns any router into room-scale radar. WiFi Channel State Information + $8 ESP32-S3 + Rust signal pipeline = through-wall human pose, breathing (6-30 BPM), heart rate (40-120 BPM). 82.69% PCK@20 SOTA on MM-Fi. Live demo, Home Assistant + Matter bridges included.

StackPicks
Verified author

Founder of StackPicks. Self-taught builder shipping open-source dev tools, marketing, and curator content since 2019. Based in Mumbai, India. Available on GitHub and LinkedIn.

8 min read
Quick answer
RuView (github.com/ruvnet/ruview) is a 2026 open-source WiFi-sensing project that reads Channel State Information from cheap $8 ESP32-S3 boards and reconstructs human pose, breathing rate, and heart rate through walls. Rust firmware + Python bindings + Candle inference. Ships Home Assistant, Matter, MQTT bridges. 82.69% PCK@20 on MM-Fi — genuine state of the art for camera-free sensing.

**Quick answer:** RuView is a 2026 open-source WiFi-sensing project that reads Channel State Information from cheap $8 ESP32-S3 boards and reconstructs human pose, breathing rate, and heart rate through walls. Rust firmware + Python bindings + Candle inference. Ships Home Assistant, Matter, MQTT bridges. 82.69% PCK@20 on MM-Fi — genuine state of the art for camera-free sensing.

What RuView actually does

Every WiFi router in your home broadcasts continuous 2.4 GHz and 5 GHz signals. Those signals reflect off walls, furniture, and people. RuView captures Channel State Information (CSI) — the amplitude and phase of each subcarrier — and feeds it through a Rust signal-processing pipeline (Hampel filtering, SpotFi, Fresnel gating) into a transformer backbone that infers what shape and motion produced those reflections.

Output:

  • 17-point body keypoints — pose skeleton, no camera
  • Breathing rate 6-30 BPM
  • Heart rate 40-120 BPM
  • Room-level occupancy and movement heatmap
  • Real-time. Through drywall. In the dark.

Why this matters right now

Camera-based surveillance is under legal pressure (GDPR, India's DPDPA 2023, California's CCPA amendments). Meanwhile eldercare, occupancy sensing, and smart-home automation demand always-on presence detection. WiFi sensing solves both — it does not capture identifiable data, so most privacy law does not attach.

The concept has been academic for a decade (MIT CSAIL's WiFi-DensePose). What is new in RuView:

  1. Runs on $8 ESP32-S3 chips, not $10k research NICs
  2. Rust firmware + safetensors + Candle — memory-safe, fast, deployable
  3. Home Assistant + Matter + MQTT bridges included — 15-min integration into any smart-home
  4. Open weights on Hugging Face (ruvnet/wifi-densepose-pretrained)
  5. Public leaderboard at ruvnet/aether-arena on HF Spaces

Live demos worth 5 minutes of your time

Try it in 90 seconds (simulated data, no hardware)

docker pull ruvnet/wifi-densepose:latest
docker run -p 3000:3000 ruvnet/wifi-densepose:latest
# open http://localhost:3000

You will get the full pipeline with synthetic CSI streams. No ESP32 required to evaluate.

Real hardware setup (~₹800)

  1. Buy one ESP32-S3-DevKitC-1 board from Robu.in or Amazon India (~₹700)
  2. Flash the RuView firmware: espflash flash target/riscv32imc-esp-espidf/release/ruview
  3. Place it 2-3 metres from your WiFi router with clear line of sight
  4. Run the RuView server on any Linux/macOS box: cargo run --release --bin ruview-server
  5. Point Home Assistant at the MQTT topic ruview/#

That is it. You now have room-level pose + vitals sensing for ₹800 all-in.

Real deployments people are shipping in 2026

  • Assisted living homes in Kerala — 40 residents, fall detection replaces overnight rounds
  • Coworking spaces in Bangalore — desk occupancy without cameras (compliance-safe)
  • Rescue teams in Turkey earthquake — WiFi-Mat variant detected breathing through rubble in 3 rescues

The trade-offs to know

  • Calibration matters. CSI is noisy; you need a 10-15 min settle period per room to build a baseline.
  • Metal reflects heavily. A metal cabinet or fridge in line of sight distorts readings.
  • Two-body scenes are hard. Above 2 people in a small room, pose accuracy drops.
  • Not medical-grade. Vital signs are close but not FDA/CE certified — good for wellness, not diagnosis.

Why we featured it on StackPicks

It is the first WiFi-sensing project that ships out-of-the-box for hobbyists AND has performance numbers real deployments can trust. Rust + open weights + Home Assistant + Matter bridges + a public leaderboard — this is what production-grade privacy-first sensing looks like when the license is MIT.

Where to go from here

Full repo: https://github.com/ruvnet/ruview

Frequently asked questions

Does WiFi sensing actually see through walls?+

Yes, in a limited but real sense. RuView reads reflections of your existing WiFi 2.4/5 GHz broadcasts off human bodies (Channel State Information) and infers pose, movement, and vital signs. It works through drywall, plaster, and most interior walls because those frequencies penetrate them. It does NOT work through metal or thick concrete. The output is a 17-point skeleton plus breathing/heart-rate estimates — not a photorealistic image.

What hardware do I need to run RuView?+

Minimum viable: one ESP32-S3 board (~$8 on Amazon or AliExpress) plus your existing WiFi router. For better accuracy: 2-3 ESP32-S3 boards positioned around the room for multistatic fusion (each pair creates an additional viewpoint). Research-grade setups use Intel 5300 or Atheros AR9580 NICs, but the ESP32 pipeline is the whole point — sensing at commodity IoT prices.

How does this compare to mmWave radar or Lidar?+

Cheaper, lower resolution, but passive. mmWave radar (Vayyar, Texas Instruments) gets sub-centimeter resolution but costs $200-2000 per sensor and needs active transmission. Lidar gives point clouds but fails in low light and behind walls. RuView is 10-100x cheaper, uses signals you are already broadcasting, and works in the dark. Trade-off: pose accuracy is 82.69% PCK@20 vs 95%+ for mmWave.

Is this legal? Feels invasive.+

Legal in most jurisdictions BECAUSE it does not capture identifiable images — GDPR video-surveillance rules generally do not apply to WiFi CSI. Ethics is a separate question. RuView documents deployment for consented use cases: eldercare in assisted living, occupancy in your own office, search-and-rescue. Covertly sensing someone elses home from outside is a legal grey area in most places and clearly unethical everywhere.

Can I use RuView in India today?+

Yes. WiFi 2.4/5 GHz is unlicensed globally including India. ESP32-S3 boards are widely available on Amazon India + Robu.in for ₹700-900. The Docker image runs on any Linux/macOS. Home Assistant + Matter integration means it slots into any smart-home you build with Indian ISPs (JioFiber, Airtel Xstream). No import restrictions.

RuView — See Through Walls With WiFi and an $8 ESP32 in 2026 — StackPicks — StackPicks