‘Can see through walls’: Open-source WiFi tool detects human movement behind walls without cameras

Open-source WiFi DensePose, released on GitHub by developer 'ruvnet', shows that Channel State Information (CSI) plus neural networks can estimate skeletal poses and even breathing-related micro-movements through walls using specialized sensors (e.g., ESP32-S3 boards) and local Rust/Python processing. The approach could enable healthcare monitoring and search-and-rescue applications but raises material privacy and surveillance concerns that could attract regulatory attention as the technology moves from labs to hobbyist use.

Analysis

This open-source RF-sensing development is an acceleration vector for two structural themes: edge AI + ubiquitous sensing and the bifurcation of “visual” vs “non-visual” surveillance. Over the next 6–24 months expect hobbyist and startup proof-of-concepts to concentrate demand on low-cost WiFi SoCs, RF front-ends and tiny ML inference engines — a demand shock that is fragmented but material for mid-cap suppliers with high exposure to IoT modules.

Second-order winners are vendors that control firmware/driver stacks and can monetize constrained access to Channel State Information (CSI) — think chipset vendors that can gate or license API access — while losers include incumbents whose value props rely on pixel imagery (camera-centric analytics and vision-SoC specialists). A plausible commercial path is integration into clinical-monitoring and insurance telemetry over 12–36 months, creating cross-sell opportunities for remote patient monitoring vendors but also concentrated regulatory risk as privacy regimes adapt.

Immediate near-term tails: (1) software-hardeners/OS maintainers could deliberately limit CSI exposure via driver updates within months, which would blunt commoditization; (2) EU/US privacy rules could mandate opt-in tagging for RF-sensing in 12–24 months, raising compliance costs and licensing opportunities. These two mechanisms are the quickest reversals of the current open-source-driven narrative and should be monitored alongside component order flows and public firmware releases.

Key Decisions for Investors

• Long SLAB (Silicon Labs) — 12–18 month horizon. Rationale: high exposure to IoT connectivity stacks and RF front-ends for low-power edge sensing. Execution: buy SLAB on any 10–15% pullback; target +35–45% if module adoption ramps; downside -25% if regulation/driver locks reduce demand.
• Long QCOM call spread (qualcomm) — 18–24 month horizon. Rationale: incumbent WiFi/SoC supplier that can monetize CSI access and enterprise licensing. Execution: buy a 2027 Jan $140/$170 call spread (debit) to limit capital at risk; asymmetric payoff if OEMs adopt licensed RF-sensing APIs; loss limited to premium paid.
• Long CRWD (CrowdStrike) — 12 month horizon. Rationale: rising attach-rate for IoT/RF device security and enterprise privacy compliance spending. Execution: buy CRWD 12–18 month calls or 6–12 month out-of-the-money calls as a conviction play; expected return 2–3x premium if regulatory-driven security budgets accelerate; risk is slower corporate capex.
• Tactical short AMBA (Ambarella) — 6–12 month horizon, small sizing. Rationale: vision-SoC vendor exposed to camera replacement risk if RF sensing displaces some camera use-cases. Execution: buy Sep-2026 puts or short 1–2% notional of book; asymmetric short: limited exposure size with catalyst being a measurable decline in camera module RFPs or design wins.