AI on the Edge Device is a firmware application for ESP32-CAM microcontrollers that digitizes analog meters by capturing images and extracting readings using on-device machine learning.
The project solves the problem of integrating legacy analog meters—water, gas, electricity, and others—into smart home and monitoring systems without replacing the meters themselves. It works by periodically photographing a meter display, extracting regions of interest from the image through inline image processing, and running those regions through TensorFlow Lite models to extract the numeric reading. The device then makes this data available through multiple output channels: MQTT, InfluxDB, or a REST API.
The tool suits anyone with non-digital meters who wants to avoid expensive meter replacement or complex manual reading workflows. It is particularly valuable for home automation integrations, especially with Home Assistant. The hardware is inexpensive and compact, requiring only an ESP32 board with a supported camera module, and setup involves flashing firmware to the device and configuring it through a web interface. The project includes OTA update capability, so firmware updates can be deployed directly through the web interface without physical access to the device.
Development activity shows consistent engagement with the codebase, with regular updates addressing functionality and user-reported issues. The project maintains active documentation and provides multiple channels for user support and feedback. Contributors have added features and improvements across the core firmware, integration points, and documentation over time.
