Unit-T & Deepomatic: Supporting the digitalization of field operations in the energy sector thanks to computer vision

MARKET : Belgium
TURNOVER in 2021 : EUR 110 million
  • KEY RESULT #1: from 15% of installations audited originally to 100%
  • KEY RESULT #2: Repeat visit rate down to 0
  • KEY RESULT #3: x5 meter installations, without expanding the back-office team
  • KEY RESULT #4: -25% time spent by technician doing reporting
  • KEY RESULT #5: Photo conformity 98%


After the European Commission set a directive to accelerate the deployment of smart meters, it was decided in Flanders that all classic meters should be replaced by digital energy meters by July 2029. This is why Fluvius, the Flemish grid company, contracted Unit-T to help it in its objective to deploy millions of smart meters for electricity and gas, replacing around 40% of its fleet of analogue meters.

Play Video

“We wanted to leverage AI to empower our technicians, certainly not to replace them. For Unit-T, it’s all about finding the optimum blending of human and technology. From a deployment perspective, we were impressed with the ease of IT implementation. Usually it can take a bit of time to develop a new AI model, but we are able to do it in one week with Deepomatic’s solution.”

Christophe De Clercq
Project Manager – Smart metering


To Unit-T, maintaining the quality and security of its field operations is essential to guaranteeing a high level of service to Fluvius and to be cost-effective in its operational processes. But the company realized that it was facing 2 challenges.

First, the photos captured by the technicians, which serve as proof that the work was done correctly and as an administrative back-up, were sometimes blurry. As a result, Fluvius had to ask Unit-T to send back a technician to complete the job again. Not only was this very costly for Unit-T, it was also a long process that could take up to 2 weeks (between the first and second visit).

Secondly, Unit-T was relying on people in the back-office to manually check the photos of a sample of operations (15% of installations), which means that most of them were left unverified.


In order to tackle these two challenges, Unit-T decided to partner up with Deepomatic. A first of its kind project in Belgium, the plan was to leverage AI-based technology in smart meters operations: technicians were assigned with a work order to install a smart meter and used Unit-T FSM app on their tablet to take photos of the meter.
Embedded with Deepomatic’s Visual Automation solution, the process involves going through over 25 control points (also called visual tasks) to make sure that it is correctly installed. The AI algorithm is able to perform several types of actions: verifying the presence of elements (gas/electricity meter, EAN sticker, seals, colored cards, reading displayed data on meter, etc.), reading data (EAN number, serial number, etc.) and checking that this data is valid (serial number and EAN format, etc.).
Finally, the solution provides direct feedback to the technician who can then correct potential anomalies.


With Deepomatic’s technology, Unit-T was able to quickly see many improvements in the field, including an increase in the number of compliant operations. The use of visual automation also helped to shorten the feedback loop with real-time quality control. Lastly, technicians have become more proficient at taking pictures thanks to Deepomatic’s solution: at the beginning of the project, they had to take again a picture for a specific task 50% of the time, and today this is almost no longer needed.


Unit-T is the partner of choice when it comes to the implementation and support of new technologies. Thanks to its 1,200 expert technicians, it is the front runner in the area of field services in Belgium. The company works in multiple industries, from telecommunications, IT, security and energy to smart buildings and cash register systems. Unit-T is part of the Solutions30 group, European leader in field services with 15,000 technicians and more than 75,000 service calls per day.