In-aisle interaction analysis with Computer Vision

By leveraging computer vision technologies, we can extract information related to in-store habits. By combining it with campaign information, we can generate high-value, tailored studies.

Objectives

• Extraction of customer movements, stops, and habits in the aisles.
• Heatmaps of traffic and detection.
• Correlation of information with offers and campaigns.

To optimise store usage and sales, it is necessary to understand and model customer behaviour in the store and, based on this data-driven knowledge, establish campaigns and offers to maximise results.

What we do

Using computer vision, we can determine customer movement and stops in store aisles, obtaining statistics that are combined with offer and campaign information, forming a dataset that enables detailed, continuous analysis over time.

How we do it

Initial data characterisation and collection.

We identify in-store areas, perspective, angles, and lighting.
We obtain reference video images for model training.
We identify and connect to store systems to obtain contextual data, offers, and promotions.

Model training and results storage

Using the images obtained, the algorithms are trained and tuned according to the requirements to maximise results. The data inferred from the image-processing systems is stored and transformed through ETL processes.

Analysis and visualisation.

Based on the stored information, we carry out analysis and reporting tasks.
The results obtained from the analyses are worked on in BI tools for further study, enabling data-driven decision-making and the execution of short- and long-term actions or promotions.