WiFi DensePose, developed by ruvnet, is a groundbreaking system that leverages commodity WiFi signals to achieve real-time human pose estimation, vital sign monitoring, and presence detection, all without the need for cameras or wearable devices. This innovative approach, described as "seeing through walls with WiFi," analyzes Channel State Information (CSI) disturbances caused by human movement to reconstruct body position, breathing rate, and heartbeat. The system utilizes a combination of physics-based signal processing and machine learning techniques to achieve its functionality.

The core functionality of WiFi DensePose revolves around its ability to extract meaningful data from WiFi signals. It achieves pose estimation by analyzing the amplitude and phase of CSI subcarriers, mapping them to DensePose UV maps. It detects breathing rates by analyzing the frequency spectrum of the signal, specifically looking for peaks in the 0.1-0.5 Hz band. Heart rate detection is similarly achieved by analyzing the 0.8-2.0 Hz band. Presence sensing is determined by analyzing RSSI variance and motion band power, providing near-instantaneous results. The system is also capable of through-wall sensing, utilizing Fresnel zone geometry and multipath modeling to detect activity up to a depth of 5 meters.

The system's architecture is designed for both performance and ease of deployment. The core processing pipeline is written in Rust, enabling high frame rates (54,000 frames per second) and a compact Docker image size (132 MB). A simple command-line interface allows for quick setup and deployment, enabling users to begin sensing within seconds. The system is designed to work with CSI-capable hardware, such as the ESP32-S3 microcontroller or research-grade NICs. While consumer-grade WiFi can provide RSSI-based presence detection, the full functionality of WiFi DensePose, including pose estimation and vital sign monitoring, requires CSI data.

WiFi DensePose offers a range of key features that distinguish it from traditional sensing methods. It prioritizes privacy by operating without cameras, eliminating the need for video data and associated privacy concerns. It provides vital sign detection, including breathing and heart rate monitoring, without the need for wearables. The system can track multiple people simultaneously, limited only by signal physics. It offers through-wall sensing capabilities, allowing it to function in environments where cameras are ineffective. The system also incorporates self-learning capabilities, allowing it to adapt to different environments without requiring labeled training data. Furthermore, the system is designed for cross-environment domain generalization, ensuring that models trained in one environment can be deployed in others.

The applications of WiFi DensePose are vast and span various sectors. In healthcare, it can be used for elderly care, patient monitoring, and emergency room triage. In retail, it can provide insights into occupancy, flow, and dwell time. In office environments, it can optimize space utilization. The system is also applicable in smart home automation, fitness and sports, childcare, event venues, stadiums, and robotics. The system's ability to penetrate walls and function in low-light conditions makes it suitable for disaster response, security applications, and extreme environments such as mining and underground operations. The system's self-learning capabilities further enhance its adaptability and utility across a wide range of use cases.