Intelligent Street Lamp Detection & Classification – Engie

Challenge

Nature of collaboration

To address this, Laborelec partnered with Sagacify to develop an AI solution capable of automatically detecting, classifying, and geolocating street lamps from video streams. The collaboration also focused on minimizing labeling costs through a custom data annotation tool and an active learning system designed to maximize efficiency.

What we built

Sagacify developed a machine learning system that processes in-car video footage to extract the location, type, and count of street lamps. The solution included:

• AI model trained for lamp detection, classification, and geolocation
• Custom labeling environment deployed for client use
• Active learning pipeline to prioritize the most impactful data for annotation

Only 700 images, labeled in under 4 hours, were required to train the final model, thanks to the data efficiency of the active learning system.

Impact