Audience Segmentation and Flow Analytics

Posted on

< Project Overview >

Introduction

The project sought to develop a hardware-software combination capable of performing demographic classification of people interacting with a vending machine. The project beneficiary is a digital advertising business. The Urban Observatory provided expert advice in development of the computer vision module. The collaboration equipped the beneficiary with means and knowledge to refine their existing product, and the Urban Observatory has discovered a potential new avenue for harvesting urban data using commercially sustainable means of advertising and purchase points.

Project aims

The project addressed IoT barriers related to costs, GDPR implications and data standardisation.

The aims were to:
– Enable and coordinate IoT image recognition hardware and an open source API
with the ability to output GDPR compliant, anonymised, platform agnostic,
audience and flow analytics data to third party applications
– Test compliance with latest GDPR regulations
– Progress and enable the sensing platforms used by the Urban Observatory with
additional IoT hardware performing demographic analysis

What was done?

Envisaged work packages included:
– proof of concept solution and detailed specification of requirements
– prototype system development on a target hardware platform
– integrated web platform for storing, access and sharing the data
Delivered work was based on existing proof of concept which has been refined and implemented on embedded compute modules (Intel NUC and Raspberry Pi). The web platform has been developed to a working prototype stage.

Results

The resulting product delivered the planned features and was able to successfully communicate with the web platform where data was displayed.

Early stage of development meant that the product has not yet been included in Hawkr’s solutions. Later perturbations related to Covid-19 prevented work in upgrading Hawkr’s fleet of digital advertising devices and no new data has been provided to the Urban Observatory. Gradual easing of lockdowns and evident turn towards more digital and less in person interactions in the physical environment is a promising tell for market adoption.

Impact

Due to little testing and no known deployment of the product its impact cannot be measured without a demo system consisting of several units being installed.. The Urban Observatory has learned about the commercial aspects of urban data harvesting.

Next steps

In order to fully understand the role and impact of the solution it is necessary to take further steps in an experimental deployment and provide necessary API endpoints for communication.

Lessons learned

  • Working at the prototyping stage presented some problems but these were overcome
  • Standardisation and privacy-awareness were fully implemented
  • A rich data platform for visualisation has been presented
  • Having a real-time issues log would help in solving technical problems at earlier stages
  • The short demo did not give full impression of what role and impact the system can have so a path to temporary deployment rather than phased refinement of the prototype would have be taken instead

It would have been useful to have applied gated access to further funding based on the quality of delivered work packages and offer incentives for taking the solution into following stages leading to market adoption. Additional resource providing oversight for such process would be welcome.

What has Pitch-In done for you?

The project has enabled the university to provide valuable technical support for our project partners and has provided us with an insight into the possibilities for capturing urban data using devices providing services for advertisers (digital screens) and customers (vending machines) which could be used in future projects.

Your cookie choices - Our website uses cookies to help give you a better experience. By accepting all cookies, you agree to the storing of cookies on your device.