This open-source project uses Channel State Information (CSI) — the subtle disturbances your body creates in WiFi signals — and feeds them through a neural network to reconstruct full human body pose in real time.
Think of it like dropping a stone into a still lake. The ripples reveal the geometry of what disturbed the water.

What makes this different:
→ Privacy-first: captures body form only, never facial features
→ Works through walls at <50ms latency
→ Tracks up to 10 people simultaneously
→ Runs on standard mesh routers you already own

The applications are staggering: fall detection for elderly care, contactless fitness tracking, intrusion detection, occupancy monitoring — all without a single camera.

Built on the DensePose From WiFi research out of Carnegie Mellon by Jiaqi Geng, Dong Huang, and Fernando De La Torre. Production-ready implementation by Reuven Cohen.

As someone working at the intersection of critical infrastructure, cyber resilience, and AI — this is exactly the kind of privacy-preserving sensing that should be embedded into smart buildings and energy systems. The surveillance paradigm is shifting.