loading

Business Impact

  • 96%

    Precision Level in Gender Detection

  • 85%

    Precision Level in Ethnicity Detection

  • 77%

    Precision Level in Age Detection

Customer Key Facts

  • Location : North America
  • Industry : Food & Beverages

Problem Context

The Coca-Cola Company, an American multinational beverage corporation, launched Coca-Cola Freestyle, their touch screen soda fountain, nationwide in 2009. The freestyle vending machines dispense over 150 different drink products and give consumers the opportunity to create their own drinks, aka “mocktails”, by mixing their desired Coca-Cola products together. Each vending machine comes with a camera installed, in which an image is captured for every customer interacting with the machine.

Coca-Cola’s Marketing team aspired to use these images to generate insights on consumer preferences and usage patterns. Therefore, Coca-Cola wanted to invest in a solution that could understand their audience as well as analyze the types of “mocktails” preferred by their consumers.

Challenges

 

  • Procurement and labeling of data
  • Real-time extraction of metadata tags
  • On-premise deployment
  • No predefined models
Challenges

Technologies Used

Python
OpenCV
TensorFlow
Convolutional Neural Networks

Accessing Real-Time Consumer Demographic Details from Pictures Captured at Coca-Cola’s Freestyle Vending Machines

Solution

Quantiphi built a custom machine learning model trained on 8 to 10 thousand images, capable of detecting customer demographics ( i.e. Age, Gender, and Ethnicity) using facial feature recognition.

As a result, Coca-Cola’s Marketing team is able to capture and evaluate demographic details of consumers, which are then used to develop focused marketing strategies and launch potential individual products, consequently driving growth.

Result

  • Smarter insights
  • Enhanced marketing strategy
  • Improved consumer experience

Looking for similar project?

Let's Talk

Get your digital transformation started

Let's Talk