City residents will take part in an experiment to create an online service using artificial neural networks. Experts from the City Department of Information Technology are working on an algorithm to more quickly relay water meter data. They will “train” the new service to automatically record water meter data from photos.
First, the neural network will process several thousand photos of hot-and-cold water meters, and it will start recording and transmitting water meter data before the year is out. The department’s officials are asking city residents to join the experiment by sending photos with water meter data to the project’s website. No registration is required.
“We are using the crowdsourcing method to involve the public in addressing various issues and tasks. This makes it possible to collect ideas and create new and useful services with the help of city residents. At least 10,000 water meter photos are needed to ‘educate’ and ‘train’ the neural network. It would be impossible to compile this impressive database without the public’s help,” said Andrei Belozyorov, adviser to the department’s head for strategic projects and innovations.
A “well-educated” neural network will be able to detect water meter data on any photo, just like the human eye does. It will accomplish this, regardless of the photo quality, resolution, lighting level or camera angle. The system will be shown additional photos in case of high error margins.
This neural network can be used to create a service for automatically identifying and transmitting water meter data to the Integrated Information and Calculation Centre whose staff will compile the utility bills, Mr Belozyorov noted.
Over 50,000 city services and website users will be invited to join the experiment. Neural networks have become a component part of software used on remote controlled cars, face-identification systems and popular photo-editing apps with various filters.
Moscow is impressively ranks among cities working to develop digital technologies. Today, almost every local family uses a city-level website for online government services.