Search

Logo Recognition Engine

Context & Motivation

Allianz is a global company with 70 OEs and offices in 70 countries and aims for a consistent usage of brand properties like the Allianz logo or specific logos for partnerships (e.g. the Olympic & Paralympic Movement OPM) around the world.
 
So it can be a challenge to perform activation measurement (how often and where is a certain logo used?) and consistency assessment (is the logo used in a consistent & compliant manner?), especially as manual assessments come at a high cost and bear the risk of inaccuracies and outdated information.

Solution

We have developed the Logo Recognition Engine, a highly automated solution to monitor & report the activation of Allianz & OPM brand properties across digital communication channels worldwide. 

The application is...

Digital

Only Cloud

Scalable

Data- Driven

How it works
The Logo Recognition Engine is based on a web crawler and a neural network that needs to be trained iteratively with the logos that are supposed to be recognized.
 
After the training phase, the Logo Recognition Engine can be scheduled to run regularly (for example every week or every month) and generate a visual report, showing which logos have been found in which location.
 
Furthermore, the tool offers the option to give manual feedback reg. logo compliance.
Result
The Logo recognition Engine has been trained to detect Allianz compliant as well as non-compliant logos literally from any abjects regardless it is a mug, a flag, a moving boat a moving sports car.
Potential Use Cases
Group Strategy, Marketing & Distribution (GSMD) wanted to find all non-compliant Allianz logos on Allianz-owned websites, tied agent homepages and social media profiles (more than 75 touchpoints of 70 OEs and Global Lines) to be able to take actions to ensure a consistent Brand identity.
Process
  • A Web crawler regularly scans all listed channels for new images
  • New images get analyzed with the Logo Recognition Engine to identify Allianz Logos
  • The Global Brand & Marketing team regularly checks all new images in the User Interface and provides manual feedback regarding the compliance of AZ logos
  • Reporting summarizes all findings on reports per channel / OE
  • Global Brand & Marketing Team uses the reports to take actions to ensure compliant usage of AZ logos
The OPM team wanted to get actionable insights into the reach & compliance of all kinds of OPM logos on Allianz-owned websites, tied agent homepages as well as social media profiles.
Process
  • A webcrawler regularly scans all listed channels for new images
  • New images get analyzed with the Logo Recognition Engine to identify Olympic Rings & Paralympic Agitos
  • Activation Operations regularly checks all new images in the User Interface and provides manual feedback regarding OPM logo type & compliance
  • Reporting summarizes all findings on reports per channel / OE
  • OPM team uses resulting reports to take actions to

           ➜  Improve reach of OPM logos

           ➜  Ensure compliant usage of OPM logos

The OPM Activation Operations team wanted to check the usage of OPM logos in OE videos that have been submitted for approval.
Process
  • OE submits a video for approval
  • Activation operations uploads the video to the Logo Recognition Engine
  • Video gets analyzed with the trained AI model to generate an annotated video showing the usage of Olympic Rings & Paralympic Agitos
  • Activation Operations uses annotated video to check compliance of OPM logo usage and informs the OE depending on the results

        

Next Steps
What you have seen so far was just a single use case of the application. With proper training, the system can detect any object from a video or image. Therefore the possibilities are endless.
 
To know more about the Logo Recognition engine and upcoming updates please contact Pablo Prieto Montes de Oca.