"The results and maturity of eSmart’s Grid Vision® solution would be very valuable to our business. The use of artificial intelligence is a great opportunity to improve our overhead lines diagnostic process, in particular by saving time and optimizing costs. Enedis is working to operationalize AI for this use case and many others.”

– Stéphane DODÉ, Project Manager, image recognition by AI, Enedis

​In a nutshell

eSmart Systems and Enedis used a collaborative approach between Artificial Intelligence and Subject Matter Experts to accelerate defect identification, increase detection accuracy, reduce costs, and improve the end-to-end inspection process.

The challenge

Historically, inspection has been mostly a manual procedure involving aerial photographs taken by subcontractors from helicopters. For Enedis, which operates the largest power distribution network in Europe, inspecting the grid means analyzing up to 36 million pictures per year (in case of a fully automatic shooting process). The process is time consuming, costly, and inaccurate.

In order to accelerate defect identification, increase detection accuracy, reduce costs, and improve the end-to-end inspection process, Enedis launched an “Automated Image Analysis System” challenge. eSmart Systems was one of the four leading companies chosen to participate in the challenge.

The objectives of the challenge were: first, to test the feasibility and maturity of AI-enabled technologies and how they can improve the whole inspection process in terms of efficiency and speed. Second, to identify potential future partners who could assist Enedis in the yearly inspection of 100,000 km of overhead lines.

Our customized solution

In order to match Enedis’ specific needs, customized AI models were developed and trained. Approximately 14,500 annotated pictures were initially made available by Enedis to train the models. The customized models were then utilized to automatically detect the relevant components and defects.  
eSmart Systems also delivered a visual interface displaying the regions where the pictures originated from as well as the entire French grid.


Our solution ranked as the #1 performing system for defect detection, both in terms of accuracy and speed. The expertise of the Subject Matter Experts was gradually captured and transferred from human to machine thanks to the annotated pictures.

The use of Grid Vision resulted in a drastic reduction in time spent reviewing inspection data, and in a better detection accuracy. 




For the largest DSO in Europe, these improvements represent a huge opportunity to rapidly enhance inspection processes and reduce costs.

Enedis was impressed by the obtained results and the maturity of AI-enabled technologies for power lines inspection. They are now exploring different opportunities to implement this type of technology into their business processes.

Want to achieve similar results?

Our customer success team ensures that our software has maximum impact and implementation across our customer base. For this reason, we have designed a 3-step customer engagement journey where we work together to understand your needs and design a customized solution for you.



1-2 days

Provides insight about our software solution and capabilities. Together, we assess your current business processes and agree on use case(s) and prioritize for maximum business impact. Next, we define criteria for converting pilot to SaaS.


2-4 months

Develop and demonstrate agreed upon use case(s). Quantify value and evaluate the solution.



Implementing the product and forming a long-term partnership. Defining and agreeing upon new use cases to ensure maximum business impact.

About the customer 

Enedis manages the public electricity distribution network for 95% of continental France, with a total of 36 million customers. Every day, its 38,691 employees oversee the operation, maintenance, and development of nearly 1.366 million km of electrical infrastructure.

To sustainably ensure high-quality grid management and reliability, Enedis performs a complete review of its overhead Medium Voltage (MV) network every 3-years, which in turn requires a yearly inspection of around 100,000 km of overhead lines. 

Take the next step 

Speak with one our experts to learn more about our product