5730

A universal, contactless smart metering solution powered by the STM32N6 with integrated NPU and camera. Using computer vision, our system can read water, gas, and energy meters without physical interfaces, offering flexibility, compliance, and rapid adaptability across different meter types.

Utility meters often require specific hardware interfaces for remote reading, such as magnetic probes or proprietary add-ons. However, in many cases direct interfacing with the meter is restricted, impractical, or demands unique hardware designs for each model. Our project introduces a contactless smart meter reading system based on the new STM32N6 development kit, which combines a high-performance microcontroller with an embedded Neural Processing Unit (NPU) and camera interface.

By leveraging computer vision directly on the device, the system can read mechanical counters, dials, or digital displays of utility meters without any physical connection. This brings several advantages:

  • Universality: The same hardware setup can be adapted to different meter models (water, gas, electricity) simply by retraining or reconfiguring the vision model.

  • Non-invasive installation: No modifications or attachments to the meter are required, which is particularly valuable in regulated environments where tampering is not allowed.

  • On-device intelligence: Thanks to the STM32N6’s integrated NPU, image capture and recognition are performed locally with low power consumption and no need for cloud processing.

  • Rapid prototyping: With the provided development kit, we built a working prototype using a flexible camera mount, a water meter, and real-time display of the recognized digits on the touchscreen interface.

The system demonstrates how embedded AI can transform legacy infrastructure into smart, connected devices in a cost-effective and scalable way. Beyond utilities, the same approach can be extended to other domains that require non-intrusive reading of instruments, displays, or mechanical counters.

This project highlights how STM32N6 empowers developers to deploy edge AI solutions that are practical, efficient, and versatile—bringing intelligence to everyday devices with minimal integration effort.