francescopace/espectre

Features

• Wi-Fi CSI Sensing - Identifies human presence and movement through walls by analyzing Wi-Fi Channel State Information without using cameras.
• Edge Inference Engines - Executes compact neural network classifiers directly on ESP32 microcontrollers for real-time gesture and activity detection.
• Motion Detection - Identifies movement via wireless signals using adjustable thresholds and automatic calibration.
• Edge Inference - Executes compact feedforward neural networks directly on the microcontroller for real-time classification of motion features.
• Motion Classifiers - Classifies statistical turbulence features using a neural network to detect human motion.
• On-Device Classifiers - Runs a compact neural network classifier directly on the microcontroller for real-time motion detection.
• Through-Wall Sensing - Identifies movement through physical barriers using wireless signals without the need for cameras.
• Wireless Motion Detection - Analyzes Wi-Fi Channel State Information to detect human presence and movement through walls without cameras or wearables.
• Wireless Sensing Toolkits - Offers an environment for prototyping and deploying wireless signal processing algorithms on microcontrollers without manual compilation.
• Home Automation Controller Integrations - Connects to Home Assistant to discover sensors and expose motion entities for automation.
• Home Automation - Integrates wireless motion sensors with smart home controllers via MQTT for presence-based automation.
• Edge ML Frameworks - Provides a runtime for executing neural network classifiers directly on microcontrollers for real-time gesture and activity detection.
• Sensing Prototyping Environments - Provides a MicroPython runtime for prototyping and deploying wireless sensing algorithms without manual compilation.
• WiFi Spatial Sensing - Summarizes per-packet signal variation into a single scalar value to measure spatial turbulence.
• CSI Metadata Extraction - Captures Wi-Fi amplitude and phase metadata from packets to monitor environmental changes via signal reflections.
• Microcontroller Runtimes - Utilizes a MicroPython runtime on ESP32 microcontrollers to enable rapid deployment of sensing logic.
• Hampel Filter Pipelines - Removes outlier spikes using median absolute deviation and applies a Butterworth low-pass filter to smooth signals.
• Data Labeling Platforms - Implements workflows for gathering and labeling wireless signal data to create training datasets for supervised machine learning.
• Signal Analysis Tools - Provides a utility for capturing, labeling, and streaming raw subcarrier amplitude and phase data for machine learning research.
• Automation Triggers - Triggers smart home actions, such as lighting or heating, based on detected room occupancy.
• MQTT Integrations - Uses MQTT messaging to adjust detection parameters and receive sensor data at runtime across devices.
• Wireless Signal Streaming - Sends real-time wireless signal data from the device to a specific IP address via UDP for external analysis.
• Firmware Flashing - Provides a command line tool for flashing compiled firmware binaries to ESP32 microcontrollers with automated chip detection.
• Hardware Signal Calibration - Provides boot-time calibration of automatic gain control and noise floors to stabilize wireless measurements.
• Microcontroller Firmware Frameworks - Interprets Python bytecode on the microcontroller, enabling rapid prototyping of sensing logic without compilation.
• Motion Sensitivity Calibration - Allows users to adjust trigger thresholds to define the level of movement required for a detection event.
• Gain and Filter Control - Locks automatic gain control and FFT gain values to ensure wireless signal stability.
• Wireless Signal Recorders - Captures and records raw Wi-Fi subcarrier amplitude and phase data to build training datasets for machine learning.
• Sensor Noise Filtering - Removes outliers and applies low-pass smoothing to clean up motion detection data.
• Sliding Window Variance - Monitors turbulence variance over a sliding window to identify motion via moving variance.
• Sensing - Provides a Python environment to develop and test motion detection algorithms before production deployment.
• Moving Variance Analysis - Detects motion events by computing the variance of turbulence over a sliding window against an adaptive threshold.
• Signal Filtering Pipelines - Implements a Hampel-Butterworth pipeline to remove outlier spikes and smooth turbulence signals.
• Subcarrier Selection Algorithms - Selects the most sensitive subcarriers using the NBVI algorithm to maximize detection reliability.
• Privacy-Preserving Sensing - Detects room occupancy using wireless signal reflections to enable automation without capturing video or audio data.
• Runtime Parameter Configurators - Supports adjusting detection settings at runtime by sending JSON commands over MQTT.
• Web Dashboards - Provides a web-based browser interface for visualizing movement, thresholds, and device statistics.

Star history