Build a Shopper Gaze Monitor

Monitor customer expressions and reactions to product positioning or advertising collateral that is positioned on retail shelves.

Target Operating System Ubuntu* 16.04 LTS
Time to Complete 45 minutes

GitHub* (C++)   GitHub (Python*)

What You Will Learn

Using a retail shelf-mounted camera system, this application counts the number of passersby that look toward the display compared to the number of people that pass without looking. It is intended to provide real-world marketing statistics for in-store shelf-space advertising.

Gain insight into the following solutions:

  • Computer vision applications for IoT
  • Inference to analyze datasets
  • Retail market IoT

Use the skills learned in this reference implementation to develop similar IoT solutions.

Learn to build and run an application with these capabilities:

Track users' expressions in response to product or market messaging.
Track statistics on customer interaction.
Interpret data from either a live webcam or preexisting video.

How It Works

The application uses a video source (such as a camera) to grab frames, and then uses two different deep neural networks (DNN) to process the data.

  1. The first network looks for faces. If successful, each face is counted as a shopper.
  2. A second neural network detects the head pose for each detected face. If the person's head is facing the camera, it is counted as a looker.
  3. The data can then optionally be sent to an MQTT machine-to-machine messaging server as part of a retail data analytics system.

The DNN models are optimized and are part of the Intel® Distribution of OpenVINO™ toolkit.

flow chart graphic of how the gaze monitor application works