Retail · Case study
Anonymous camera observations were connected to spatial movement and shelf-attention signals.
Case study context
A retailer wanted to understand how shoppers moved through a physical store and which shelf areas attracted attention before a purchase, without identifying individuals. Sales records showed only the final transaction, while people counting and person detection could not reveal what someone looked at. An ordinary store camera also provided no reliable training label for the shopper's true gaze direction.
The store was mapped into a shared coordinate system and anonymous camera tracks were localized within it. During model development, eye-tracking glasses supplied the missing gaze ground truth, linking camera imagery, shopper position and the shelf actually viewed; deployment then required only the ordinary camera feed to estimate shelf-attention events.
Attention evidence thread
Use the controls below to inspect the key stages, constraints and validation decisions in that process.
Move through a short example to see how a shopper position becomes a shelf-level attention pattern.
Evidence retained: Review connected the source camera frame, anonymous position, estimated gaze target and accumulated shelf attention.
Start with the problem
Share the environment, constraint and result you need. We can discuss the engineering path without exposing confidential details.
