Retail Customer path analysis on Google Cloud

One of Australia’s leading entertainment providers wanted to optimise customer sales in their stores for many locations across Australia. Vision and video analysis using Google Cloud Platform provided a unique opportunity for Servian to provide insights on individual shopper’s paths for the client.

problems & pain points

Servian was engaged with to provide a way to improve customer interactions using Machine Learning while leveraging Google Cloud capabilities and existing video footage. The goal was to optimise customer sales and marketing through leveraging insights on individual shopper’s paths. The primary challenges were:

outcomes and results

Servian created a video analysis code for the client that would run across GCP and allow for testing on viable machine learning techniques. To achieve the outcomes required, Servian constructed Dockerised solutions on GCP that utilised Google Cloud Storage and Google Compute Engine.

The work undertaken by Servian on GCP achieved time based insights from their store video analytics as well as reporting on solution pros and cons. This included time based analysis in Python of video surveillance, including visits and checkout decreases over time. In addition the development of object tracking mechanism and of custom point of interest event detection algorithms allowed for detection of where the shoppers spent the most time inside the location. Servian used batch processing of sample videos supplied with Python data analytics from video processing on GCP to show product popularity as well as congestion and dwell time inside the client locations.

There are many more applications with this kind of data, one area of possible improvement shown from the video analytics is optimal locations for future cameras to capture important in store activity.

platform used

To achieve the outcomes required, the following GCP products were used: