Reliable electricity supply is paramount to Caruna. As their electrical network spans multiple regions, supervision and maintenance is time-consuming, particularly in rural areas. Caruna had developed a system to photograph and monitor their electricity distribution components, but manually processing thousands of photographs was laborious. Caruna approached us with the task of creating a computer vision algorithm that could accelerate supervision and maintenance.
We developed and trained a computer vision algorithm to analyze images of electricity distribution components and pre-process them for abnormalities / faults and categorize them based on their types. After a successful pilot, we designed and created an application that Caruna’s maintenance team now uses to support their work.
- Custom solution, comprising a computer vision based object detection module and a rule-based anomaly detection module
- Modern web app user interface for ease of access and control, deployed on cloud infrastructure
The application has decreased the amount of images that Caruna’s employees need to manually process by 50%. As more material is collected, the algorithm and application can be developed further to streamline this process or add new use cases.