Computer vision is a relatively untapped technique that offers marketers significant ways to ensure brand safety. Ed Preedy, MD Europe at applied computer vision company, GumGum, talks about why the tech has become so crucial for advertisers.
It is said that a week is a long time in politics, but sometimes just a few months can feel like a lifetime in marketing. In just under a year, the threat of online brand safety has become the top issue keeping those responsible for brands awake at night.
Whether it’s the horror of discovering that ads for your family-oriented brand have appeared next to terrorist videos, or the embarrassment of finding your car ads surrounding graphic pornographic content, it seems that all the benefits of programmatic media buying have also ushered in some huge risks to brand reputation.
Now new research from GumGum has given us some more detail on exactly what is most feared, what is actually happening and what marketers are doing about it. In many cases we found that they are missing one technique that could prove highly useful – just because of lack of knowledge.
The fear, it seems, revolves primarily around the possibility of brands being associated with hate speech, pornography and violence. Hate speech was named by 34% of respondents, while the latter two issues worried 17% and 13% of marketers respectively.
The fear, it seems, revolves primarily around the possibility of brands being associated with hate speech, pornography and violence. Hate speech was named by 34% of respondents, while the latter two issues worried 17% and 13% of marketers respectively.
When our research looked at how marketers are trying to tackle these risks we found that keyword detection is still the first port of call for most marketers trying to keep their brands safe – 53% use this. Drawing up blacklists of sites likely to host bad content is also popular with half of marketers.
But relatively few marketers are using any computer vision-aided techniques to help their brand remain safe. Otherwise known as image recognition technology, this refers to the ability of a machine to receive, analyze and make decisions based on visual data like photos, video or sensor data.
At a basic level it is the technology that allows your smartphone camera to work out where the faces are and focus on them. It is also helping to power driverless cars.
44% of marketers have come across dubious imagery sitting next to their brand’s communications, with 32% having dealt with unsavory video footage in similar contexts.
For brand safety, the technology has massive benefits. So much of what constitutes dangerous content on the internet is visual, from pornography to video footage by extremists. This is also borne out by our research – 44% of marketers have come across dubious imagery sitting next to their brand’s communications, with 32% having dealt with unsavory video footage in similar contexts.
But despite this, relatively few are employing computer-vision techniques to combat the problem. Just 15% scan pre-placement, with 11% using it as a corrective. Publishers are even more in the dark – only 2% use computer vision to clean up their own content.
This is partly down to simple lack of knowledge. It is only in recent years that the internet has become so drenched in imagery. Marketers are still text-obsessed because keyword recognition is one of the oldest ways of attempting to control the online environment.
But the reality is that unsafe online imagery can often appear without any accompanying words at all, meaning that it is only a technique like the AI-powered computer vision that can hope to detect it. At GumGum, as well as using our computer vision technology to provide hyper-relevant targeting for brands, we can also scan the internet for images like guns, swastikas and nudity, offering marketers a way of avoiding dangerous content that keyword detection and sometimes even blacklists would be unable to help with.
Computer vision has a significant role to play in a wide range of industries, from healthcare to security. Arguable though, one of the most compelling areas is the role it could play in helping to make the internet safer for both brands and consumers. That’s why we believe all marketers will eventually employ computer vision as a significant part of their brand safety efforts.