AI-On-The-Edge-Cam is an ESP32-S3 board with a 2MP camera, microSD card, WiFi, BLE, and PoE Ethernet connectivity designed to digitize legacy utility meters such as water meters, gas meters, or electricity meters that require manual onsite reading.
AI-On-The-Edge-Cam is also the name of the firmware running on the board and was initially designed for the ESP32-CAM board. But the new board adds a few features that make it more versatile, like PoE support and a microSD card slot for storage, as well as RGB LEDs acting as a flash to allow meter reading at any time of the day or night.
AI-On-The-Edge-Cam specifications:
- Wireless module – ESP32-S3-WROOM-1U-N16R8
- SoC – Espressif Systems ESP32-S3
- CPU – Dual-core Tensilica LX7 up to 240 MHz
- Memory – 512KB SRAM, up to 8MB PSRAM
- Storage – 16 MB flash
- Wireless – WiFi 4 and Bluetooth LE 5
- IPEX antenna connector
- SoC – Espressif Systems ESP32-S3
- Storage – microSD card slot
- Camera – 2MP OV2640 camera module
- Connectivity
- 10/100 Mbps Ethernet RJ45 port with PoE
- WiFi and Bluetooth on the ESP32-S3 module with IPEX antenna
- Expansion
- Stemma QT connector
- 2x 6-pin headers with I2C, UART, GPIO, 5V, 3.3V, and GND
- 3-pin headers with LED control pin, 5V, and GND
- Misc – WS2812B back-lighting LEDs
- Power Management
- 5V via USB-C port
- PoE
- Battery management
- Low-power sleep mode drawing about 20 µA
- Dimensions – TBD
The AI-On-The-Edge-Cam board ships with a 2.4 GHz antenna and a 2MP OV2640 camera module. The firmware, hosted on GitHub, supports Tensorflow Lite (TFLite) integration, inline image processing, a Web interface for control, OTA firmware updates, Influx DB 1 and 2 databases, integration with Home Assistant through the MQTT protocol, and a REST API.
It can read the large digit on the meter as well as the small dials for digits after the decimal. Once it has the data, it can store it in a database on a microSD card and/or send it to a gateway via MQTT or the REST API. The firmware is not quite new, as it was released about 5 years ago and is still regularly updated with an active community. The video below shows how to use the ESP32-CAM board with Home Assistant.
The AI-On-The-Edge-CAM board, however, appears to have just been introduced and had two pieces left in stock selling for $28 on Tindie when I started this article, and is now out of stock.
You can also use another ESP32 or ESP32-S3 board with an OV2640 camera, but you’ll have to make sure it has at least 4MB of PSRAM, or this won’t work. The XIAO ESP32S3 Sense board could be a candidate, although it only has one user LED, which may or may not provide enough illumination. The Waveshare ESP32-S3 ETH board would be another option with PoE and a microSD card slot, and (RGB) LEDs could be added through some of the I/O pins.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress. We also use affiliate links in articles to earn commissions if you make a purchase after clicking on those links.
Been using this software for a while on the traditional esp32-cam, it’s just too sensitive to lighting and exact sizes. If you can’t get it absolutely perpendicular to the plane of the text it gets confused. If you have lighting from the environment, the variable nature screws it up, if you use the board LEDs the reflections bugger up the character recognition. You end up having a specially constructed mounting system with remote relocation of the board LED and even then the values struggle with slow moving meter reads, with tens of digit jumps common.