Get personal with logo detection
To provide customized designs to its end users, a leading marketing technology firm came to us seeking an ability to detect and extract logos automatically from sets of random brand images.
Image recognition system
Focused on improving personalization in its marketing efforts, a key client wanted the ability to extract images from random websites and detect any branding and logos in the content. The easiest way to distinguish branding from other images is through an image recognition system, which requires training data and a deep neural network, or a convolutional neural network (CNN), which is used to analyze visual imagery.
Jobs to be Done
- Extract logos and branding from websites
- Develop a large training set
- Develop neural network model
- research & development
- develop training data set
- develop neural network model
- training data set
- trained neural network
Through our machine learning capabilities, our developers created a large training data set and applied a CNN to deliver a 99% accuracy rate in image classification in our client’s use case. Through our image recognition system, we were able to identify logos and their placement, size, and prominence.
Automation increases end-user productivity
Thanks to this advanced automation, our client’s end users are now able to instantly extract logos from images, which significantly boosts user productivity and enhances the customer experience. As a result, our client received a significant “wow” response from its end users.
In addition, by automating logo identification, our system could auto-suggest logo placement and usage throughout our client’s marketing application, further enhancing the end-user experience and potentially driving increased revenue.
Ready to kick off your machine learning project?
next case study
innovation research, machine learning, media / technology