There are thousands of cases of water damage in Finland every year. LeakLook decided to tackle the challenge using a water safety service implemented harnessing the opportunities provided by AI. A modern water meter has enabled customers to identify water leaks in time and better control their daily water consumption.
Water damage is a common occurrence in Finnish homes. According to Finance Finland, on average, there are 100 cases of water damage per day in Finland, and the compensations paid for them are well in excess of EUR 100 million at the annual level. Unfortunately, water leaks often go unnoticed until the damage has already been done. Without exception, several thousands of euros are spent on repairing water damage. Therefore, it is important that all water leaks are detected as quickly as possible to avoid extensive damage and costs.
LeakLook accepted the challenge by implementing a contemporary solution for detecting leaks. Our collaboration is aimed at digitising water consumption monitoring, reporting and leak control in detached houses, housing companies and the industry. What makes our solution particularly smart and light is that it enables digitalisation without having to replace the old and functional analogue meters. The update applies to the reader, not the meter itself.
Machine vision: Ready OpenCV libraries. The structure of the neural network was designed for purpose and trained using self-collected data. The machine vision runs on Amazon cloud, where the devices send data via the mobile network.
Technologies: Python, Keras, Docker
The collaboration between Vincit and LeakLook started with improving the user experience and bringing the user interface to a level that meets the customers’ expectations. The aim was to make the deployment of the water safety service as easy as possible for the users, right from the start. The idea led to a product that everyone can easily install on top of their existing traditional water meter. The water safety service monitors the water meter readings using machine vision and transfers the consumption data securely to the user, who can read it on a desktop, mobile phone or tablet, 24/7. The customers are able to connect several sites to the service, for example, home and summer cottage. The service gives an alarm in the case of a leak or some other anomaly in water consumption.
Vincit has been involved in the development of nearly all things related to the operation and customer UI of the LeakLook product, from the integrated device system to the online platform used by the customers. With ready-made solutions proving to be unworkable, a bespoke AI was developed specifically for machine-read meters and trained for the task using freshly collected data.
The LeakLook water safety service has introduced many benefits to the users. The water meter has made it possible to predict and manage household water consumption and water bills. Vincit and LeakLook’s collaboration will continue. At the moment, we are working on gradual improvements and training the machine vision.
LeakLook is a water safety service that provides households with water consumption transparency as well as cost-effective water safety for the home. The LeakLook service comprises a water meter reader and an alarm service.
