We use neural networks - computer systems modelled on the human brain - to analyse photographs taken on site while installing IoT devices.

This intelligent analysis helps to fix any errors in data input and verify the quality of installation, ensuring the integrity of network data and enabling huge savings for our clients compared to the cost of manual auditing.

To Interpret this kind of photographic data, we must develop complex algorithms capable of:

  • adjusting for light conditions
  • adjusting for photo angle
  • recognising whether a photo contains a device at all, and if so which kind
  • understanding where to look for the same data in different devices
  • interpreting difficult-to-read interfaces such as manual meter dials.

Towards fully automated auditing

Emerging applications of this technology go well beyond reading data. For example, we can use neural network image processing to determine if a device has been installed correctly, down to the last screw.

By offering this capability to clients, we minimise their reliance on manual audits, improving accuracy rates while saving costs.