How Machine Learning Techniques Help Register and Protect Logo Marks

Published: March 17, 2021

Charles Hill Corsearch New York, New York, USA Brands and Innovation Committee—Content Subcommittee

The history of logo marks can be traced back at least as far as Ancient Greece and Turkey when monogrammed coins were used to represent city-states. The practice developed and by 1300 AD businesses were also using signage to signal their identity and represent what they did. The first modern logo designs were created in the late 1800s to capitalize on new mass printing opportunities.

Famous examples include the Samson Rope Technologies Inc. logo. Registered with the U.S. Patent and Trademark Office in Class 22 (Cordage and Fibres) in May 1884, it is still in use today (renewed in December 2013). The mark consists of a picture representing Samson and the lion on top of the stylized, partially underlined word SAMSON.

Benefits of a Good Logo

In today’s world, as soon as a logo is used in commerce it becomes the identity of the company, product, or service it represents. A good logo distinguishes the offering, helps a brand stand out from the competition, and attracts potential customers. It can be placed anywhere—from packaging and billboards to websites and social media. A recognizable logo mark helps brands to foster, build, and preserve brand loyalty.

Intelligent Logo Check

While logo registration is extremely important, the trademark search process has been notoriously challenging given the historical solutions available to practitioners—with no source having complete information and there being reported difficulties with image descriptor codes (such as Vienna codes). In the past, this process was slow and partial and required a rigorous manual search of national and international registries for similar marks.

This is where new and innovative image-to-image search solutions can help. Today, machine learning is doing the heavy lifting on much of this work. Deep learning neural networks have been rigorously trained against millions of logo image samples and can instantly break candidate logos or designs down into their component parts to compare the components’ similarities (including their shapes, concepts, letters and words, and colors) against those of existing logo images.

This kind of artificial intelligence (AI) is empowering a new generation of tools that can instantly check new logos against millions of other third-party marks. It is allowing intellectual property (IP) professionals to run more complete design searches in less time and get back many hours to focus on other work.

Intelligent Logo Protection

Machine learning is also driving positive innovations for IP professionals and brands when it comes to reducing counterfeiting and trademark infringement. Intelligent image technology has features which can recognize and label objects within any image. This is done via “computer vision,” which means teaching the AI machine what an object looks like so that it can then show you all images where it finds examples of those objects. The object can be anything that it is programmed to be, such as a brand logo or any other recognizable mark.

Brand owners can even use their own image libraries and help train the AI to recognize their logo marks. This makes it possible to use smart image matching technology in the management of brand portfolios.

Importance of Protecting Brand Logos

All IP rights are a vital driver for global economies. Industries that rely upon these rights create job opportunities, contribute to gross domestic product, and pay taxes. Counterfeiting is a bigger menace than ever, however, and brands are continually exposed to the activities of logo infringers using their IP to try and dupe consumers.

In addition to using brand logos to advertise fake goods, infringers also steal copyrighted images and use them on multiple platforms to offer counterfeits. Brand owners can counter this, however, and enforcement success rates increase when companies can monitor images containing their logos. Intelligent logo detection is critical then because it supports brands to quickly see where their logos are being infringed and take swift action.

Ultimately, it is not just the wealth of producers that is compromised by this kind of illegal activity. The health of consumers is also put at risk by counterfeit goods, many of which are produced with no regard to the health and safety of the end user.

From clearance to protection, we are now seeing how machine learning can support the safeguarding of societies and also streamline our work as IP professionals.

Charles Hill is vice president of product marketing at Corsearch in New York, New York, USA. He has over 20 years’ experience managing large-scale software development and product launches, and is a published author, industry speaker, and thought leader on AI and process innovation in IP and legal tech.

Although every effort has been made to verify the accuracy of this article, readers are urged to check independently on matters of specific concern or interest.