ESP32-S3 PoE AI Camera Board Makes Your Legacy Utility Meter Digital and IoT Enabled

AI-On-The-Edge-Cam – Image courtesy of github.com/allexoK

Designed by Prokyber in the Czech Republic, the AI-On-The-Edge-Cam is an ESP32-S3-based development board designed to digitize legacy utility meters, such as water, gas, and electricity meters that still require manual reading. It builds on the original AI-On-The-Edge-Cam firmware, while also adding PoE, RGB LEDs, and a microSD card slot.

As the board is built around an ESP32-S3 MCU, it supports WiFi 4 and BLE 5, on top of that it has support for 2MP OV2640 camera, microSD storage, and PoE-enabled Ethernet. and also includes multiple expansion headers (Stemma QT, GPIO, UART, I2C), WS2812B RGB LED, and a USB-C port for 5V power and programming. These features make this board suitable for retrofitting utility meters with edge-based OCR and remote monitoring.

AI Utility Meter Digitizer on the Edge – Image courtesy of github.com/jomjol

AI-On-The-Edge-Cam Specifications

• SoC: ESP32-S3-WROOM-1U-N16R8 wit dual-core LX7 @ up to 240 MHz
• Memory: 512KB SRAM, up to 8MB PSRAM
• Storage: 16MB flash + microSD card slot
• Wireless: WiFi 4 and Bluetooth LE 5, IPEX antenna connector
• Camera: 2MP OV2640 camera module
• Ethernet: 10/100 Mbps RJ45 port with PoE support
• Expansion:
  • Stemma QT connector
  • 2x 6-pin headers (I2C, UART, GPIO, 5V, 3.3V, GND)
  • 3-pin header (LED control, 5V, GND)
• Misc: WS2812B RGB LEDs for notifications
• Power
  •  5V via USB-C, PoE, battery support
  • ~20 µA low-power sleep mode
• Dimensions: Not Mentioned

The board runs the open-source AI-On-The-Edge-Cam firmware available on GitHub, with support for TensorFlow Lite inference, inline image processing, MQTT and REST API protocols, OTA firmware updates, and integration with Home Assistant and InfluxDB. It includes a web-based UI for setup and control.

AI-On-The-Edge-CAM – Image courtesy of github.com/allexoK

The AI-On-The-Edge-Cam firmware is capable of reading both the large numeric digits and the smaller dials representing decimal values on utility meters. After extracting the readings, the data can be stored locally on a microSD card or transmitted to a gateway using MQTT or a REST API.

Although the firmware was initially released around five years ago, it remains actively maintained and supported by a strong community. A demonstration video shows how the firmware can be used with an ESP32-CAM board and integrated into Home Assistant for automated monitoring.

The AI-On-The-Edge-CAM board is priced at $28 and is available on Prokyber’s  Tindie Store.