Turn ordinary WiFi signals into real-time vital sign monitoring, presence detection, and through-wall sensing without a single camera, wearable, or cloud service. That is what RuView does, and the open source project just crossed 84,000 GitHub stars with over 10 million downloads on its Rust crate alone.
RuView, built by the team behind RuVector and Cognitum Seed, is a WiFi sensing platform that turns commodity ESP32 microcontrollers (as cheap as $9 each) into powerful spatial intelligence sensors. Instead of adding cameras or wearables, RuView analyzes Channel State Information from existing WiFi signals to detect people through walls, measure breathing and heart rate, recognize falls, and even map room layouts. It runs entirely on edge hardware with no internet connection required.
How WiFi Sensing Actually Works
Every WiFi router already fills a room with radio waves. When people move, breathe, or sit still, they disturb those waves in measurable ways. RuView captures these disturbances using ESP32-S3 boards that cost $9 each. A small pretrained AI model processes the data in microseconds on a Raspberry Pi.
The numbers are impressive. The system detects presence in under 1 millisecond with a v2 encoder that achieves 82.3 percent held-out accuracy on temporal triplet tasks. It measures breathing rate between 6 and 30 BPM in real time, and heart rate between 40 and 120 BPM. It can even estimate 17-keypoint pose from WiFi signals alone, hitting 82.69 percent torso-PCK on the MM-Fi benchmark. That beats every previous approach, including MultiFormer at 72.25 percent and CSI2Pose at 68.41 percent.
What makes this particularly interesting is the self-supervised approach. The team trained a contrastive encoder on 60,000 frames with zero manual labeling. Each retraining epoch takes 84 seconds on an M4 Pro Mac. Any founder building a physical sensing product should study this architecture because it sidesteps the hardest problem in IoT: labeled training data at scale.
What You Can Build With It
RuView ships with 21 entities per sensor node and integrates natively with Home Assistant, Apple Home, Google Home, and Amazon Alexa. A founder could deploy RuView in under an hour with a single Docker command and get real-time room-level occupancy, sleep quality monitoring, and elderly inactivity anomaly detection.
Key capabilities include:
- Presence and occupancy tracking through walls: Detect people in adjacent rooms, count them, and track entries and exits with zero camera privacy concerns
- Contactless vital signs: Breathing rate and heart rate measured while someone sleeps or sits still, with no wearables. This works through blankets, through walls, in complete darkness
- Fall detection under 200 milliseconds: Phase-acceleration threshold with 3-frame debounce and 5-second cooldown, designed for elderly care
- Multi-person counting: Self-calibrating with adaptive normalization, supports six specialized learned counters for elevators, queues, and clean rooms
- Through-wall sensing up to 5 meters: Uses Fresnel-zone geometry with multipath modeling. It even uses your neighbors WiFi signals as free radar illuminators
- Sleep monitoring with stage classification: Overnight tracking with apnea screening, all from radio reflections
The entire system runs on edge hardware with a bill of materials as low as $140. Every measurement is cryptographically attested via an Ed25519 witness chain, and the pretrained model fits in just 8 KB when 4-bit quantized. The team published the model on Hugging Face at ruvnet/wifi-densepose-pretrained for anyone to use.
What This Means for Founders
RuView opens up categories that were previously locked behind expensive sensors or privacy-hostile cameras. Three opportunities stand out:
1. Senior care monitoring without cameras. The elderly inactivity anomaly and fall detection features address a massive and growing market. Nursing homes and assisted living facilities are desperate for monitoring solutions that do not violate resident privacy or require residents to wear devices they will forget or remove. A $9 ESP32 in every room beats a $200 camera with all the consent, compliance, and resident acceptance baggage.
2. Smart home differentiation that actually works. Every smart home platform claims presence detection. Most rely on motion sensors that miss stationary people or cameras that creep people out. RuView offers through-wall detection without line of sight and with vital signs as a bonus. Founders building on Home Assistant or Matter should integrate this the moment it reaches production stability.
3. Industrial IoT without the infrastructure cost. Through-wall occupancy, people counting in warehouses and clean rooms, and queue length tracking are directly applicable to logistics and manufacturing. The 105-cog edge module catalog covers health, security, building, and industrial use cases out of the box, and the entire stack runs on hardware you can buy at a hobby electronics shop.
There is a catch worth noting. The system requires CSI-capable hardware, which means ESP32-S3 boards or research-grade NICs for full capability. The Docker image runs with simulated data without CSI hardware, but the real value comes from physical deployment. The 82.3 percent accuracy figure is also honestly benchmarked the team retracted earlier overstated claims and published the corrected numbers with full auditability on AetherArena. That kind of transparency is rare in open source AI and earns trust.
Bottom line: RuView proves that WiFi signals are a sensor platform most of us already own. For $9 per node, you can add spatial intelligence to any space without cameras, without wearables, and without cloud fees. The 84,000 stars and 10 million downloads suggest developers agree. If you build anything in IoT, smart home, or senior care, this repo deserves a deep look today.