IT Brief UK - Technology news for CIOs & IT decision-makers
United Kingdom
Avju adopts eSmart AI Studio for Gridvju vision models

Avju adopts eSmart AI Studio for Gridvju vision models

Tue, 17th Nov 2026 (Today)
Joseph Gabriel Lagonsin
JOSEPH GABRIEL LAGONSIN News Editor

Avju Solutions has adopted eSmart Systems' AI Studio for the computer vision models used in its Gridvju platform, supporting its work across Europe and North America.

Avju focuses on dynamic line rating and uses Observer units mounted directly on live power lines to capture 4K imagery, along with electrical and environmental measurements. Utilities use the resulting data through Gridvju to monitor line capacity, asset condition and risk.

Before adopting AI Studio, Avju's data and autonomy team built its computer vision models internally. Those models performed well, but maintaining them took up much of the effort of a small team, particularly as deployments expanded into new environments that needed additional training data, labelling and retraining.

Avju now starts with a library of utility models, then builds and runs its own models within AI Studio. Its engineers also use AI-assisted development tools alongside the platform, feeding the results into Gridvju.

First use case

The first production use case is corona discharge detection on insulators. Visible corona activity can be identified in Observer imagery and combined with an insulator defect detector to help grid operators assess performance and spot developing problems earlier.

AI Studio is also being used to detect people, faces and vehicle number plates in imagery so those areas can be redacted where required. Other use cases highlighted by the companies include fire and smoke detection, tree clearance, and snow and ice detection.

The companies said adapting models to local conditions is an important issue in power networks. Infrastructure varies between countries in mast designs, insulators, materials, lighting, weather and camera angles, meaning a model trained in one setting may face unfamiliar conditions elsewhere.

Avju said the platform allows it to reuse utility-specific models, adapt them to local conditions and add new ones through the same system. That allows its computer vision work to expand with Observer deployments without needing a separate model set or team for each market.

The software approach also means new detection functions can be applied to imagery already captured by equipment in the field, without changing the installed hardware. For utilities, this can reduce the volume of raw imagery that needs review and provide earlier warning of developing faults through Gridvju, the companies said.

Utility focus

eSmart Systems said AI Studio has three components: Model Builder, for creating custom models from a small number of example images; Model Garden, a library of detectors and classifiers for utility infrastructure; and Pipeline Builder, for linking models into workflows. It added that each model and pipeline has its own API endpoint, and that models created in AI Studio can be carried into its Grid Vision platform.

According to eSmart Systems, its Grid Vision platform is used by more than 85 utilities across the United States, Canada, the UK, Ireland and continental Europe. Avju's solution is also being deployed in the United States through NATO DIANA's U.S. Market Access Program.

"AI Studio gives every team access to proven models built on tens of millions of images and a decade of eSmart experience in utility infrastructure, and those models keep improving," said Erik Åsberg, chief technology officer at eSmart Systems. "Adaptive AI then makes it easy to build and maintain new ones from a handful of images, so a small team with limited data can move at the speed of a much larger one. Avju is showing what that looks like. It changes how fast a team can innovate and how far it can scale."

Avju described the arrangement as a way to supplement a small internal team with access to established computer vision work in the utility sector.

"eSmart Systems has a super strong reputation in computer vision and extensive experience applying it to utility infrastructure. We gain more than access to technology. We gain the depth and confidence in the computer vision capabilities behind our products," said Jonas Mørch-Lampe, head of data & autonomy at Avju Solutions. "That's a foundation we couldn't realistically build alone as a small team."