Case Study Logo Detection

Logo Detection Solution

Summary

A media company based in Germany needed an automated and AI-powered solution to gauge brand visibility during broadcasted sports events and tournaments. To build this solution they partnered with Folio3 AI for a custom logo detection solution that detected brand logos in sports and tournament broadcasts uploaded on their existing platform.

About the Customer

This Client provides innovative solutions for digital Sports and Entertainment, bringing together athletes and content creators from several sports, such as football, fitness, BMX, e-sports, etc.

logo-logo detection
  • Team composition

    4 members

  • Expertise used

    Computer Vision, Machine Learning, Deep Learning

  • Duration

    15 weeks

  • Services provided

    AWS infrastructure Configuration, System Access APIs, AI Model pipeline

  • Country

    Germany

  • Industry

    Spectator Sports

Understanding the Challenge

Keeping track of all the brand logos in a sports event was quite tricky as many sponsored brands display their logos at different areas in a game. To make sure a specific brand’s logo is displayed on the screen during an event and to keep track of a brand logo visibility, they needed a solution that could detect logos and give insights on information for quick analysis and ROI calculation.

 

Solution

Folio3 AI built a solution for our client that could detect multiple brand logos in images and video clips extracted from various sports events. The model was trained and fine-tuned on the data set provided by our client using computer vision and deep learning techniques.

Customized AI model

The algorithm was fine-tuned and trained on the client data set and customized according to solution needs to detect various brand logos

customized AI model

Sponsorship Monitoring

By detecting logos at different locations in an event clip, our client was able to monitor and analyze the ROI for the event sponsors, counting how many times a logo was detected and where.

Sponsorship mentoring
logo detection 2nd
logo source detection

Logo Source Detection
 Locate logos at different locations during the event and provide results accordingly i.e. (player jersey, signage, digital graphics, floor, etc)

model trained

Model Trained on 2000+ Brands

The model was trained to detect logos of more than 200 brands from anywhere within an image or a video clip uploaded on the client platform.

Result

With seamless integration onto the client’s existing platform, the model automatically detected client logos in images and video clips uploaded on the platform, saving hours of manual work. The model also helped our client calculate the ROI for all event sponsors.

Looking for a Similar Solution?