Type to search

Brands better recognise: why you should care about image recognition

Technology & Data

Brands better recognise: why you should care about image recognition


Jon Stubley says marketers should be paying attention to image recognition and using it to understand their audiences and protect their brands.


Early in March, during the keynote of its Cloud Next conference in San Francisco, Google announced the launch of a new machine learning API for automatically recognising objects in videos and making them searchable.

The announcement is significant as videos have historically been a challenge for machine learning researchers to extract information from. At this stage its applications are yet to be defined, but it has the potential to be impactful in the world of media and marketing. As the company noted, “Google has a long history of working with the largest media companies in the world, and we help them find value from unstructured data like video.”

While its impact on the world of media and advertising is yet to come, the impact of machine learning when applied to static images for the media and advertising industry (and beyond) is already significant.

I’ve worked for GumGum since we launched in Australia one year ago and since then, the pace of change, not only in my own company, but in the broader computer vision (sometimes known as image recognition space) has been nothing short of phenomenal. Images hold a wealth of information and data that are incredibly powerful for marketers and the computer vision space is driving real transformation and innovation across industry sectors.

If you work in marketing and aren’t already considering how image recognition could help your marketing strategy then now is the time to do so. And here is some food for thought as to why.


The Snap effect

First off, the Snap Inc IPO. I mean whoa. Since launch it’s been volatile but at the time of writing it is valued at over US$28 billion. That’s considerable inflation on the old saying ‘a picture is worth a thousand words.’

Valuation aside, there is no doubt that Snapchat is reshaping digital media. It has risen to its multi-billion dollar heights by tapping into our seemingly insatiable appetite for sharing photos. Last year, Snapchat announced it had four million active users in Australia, which equates to about 25% of all smart phone users, with 150 million globally (Facebook has 12 million local daily active users, and Instagram 7 million).

Care factor?

The sheer volume of images being uploaded and shared on a daily basis should tell brands that images are the new currency of consumers. Outside of the Snapchat advertising offer itself (local brands Unilever, McDonalds, Myer, AAMI and Dan Murphy’s are just some of the users of its advertising platform), brands should consider serving image-based ads that are contextually relevant to the people viewing and sharing them.

Computer vision technology is now sophisticated enough to analyse a photo of a bedroom and serve up an ad matching a specific paint colour to it – and all in real-time.


Know your audience

Last year, a US based Facebook study of over 160,000 people used a specific form of image recognition to automatically pick up photos of users with either dogs or cats in their photos and analyse their posts.

It found that cat lovers were more likely to be single, on average dog lovers have 26 more Facebook friends than cat people, but that ‘cat people get invited to more events’ and express a ‘wider range of emotions’, including exhaustion and annoyance, and have an unusually strong interest in fantasy, anime and science fiction.

Care factor?

Shared social media photos hold a wealth of information about brand preferences, shopping habits and cultural sign-posts that are incredibly insightful, and useful, for marketers. Take the above example. In Australia, the pet food market is estimated to be worth over $3 billion a year, with 90% of the market going to dog and cat food. The potential upside of being able to identify cat and dog lovers is huge.


Protect your brand

Last year, tech start-up Trademark Vision, which was founded by former NICTA (now Data61) senior computer vision research engineer Sandra Mau, won a lucrative contract with IP Australia to transform the agency’s search systems. It launched in February this year and images can now be used to search IP Australia’s trade mark database of over 400,000 trade mark applications. Their technology can also be used by law firms and corporates to protect brands through visual search.

Care factor?

Incredibly useful for brands looking to protect their major IP assets, this type of technology can also be used by brands to keep track of how their brand is being represented across shared photos and images across devices. In the US, my own company has a visual intelligence social listening solution that is able to identify and connect to influencers who share images that relate to a brand and gain insight into how these influencers interact with images in order to inform earned media strategies and activate paid campaigns.


Saving the planet

Australian start-up Jemsoft’s Monocular API is a computer vision machine learning technology conservationists  are using to apply to drone footage from the Sumatran jungle to assist in preserving orangutans. Instead of manually reviewing photos to identify their nests, the software can be trained to find them.

Care factor?

It’s orangutans; what’s not to care about?


Interested readers should also check out US app Markable which lets people draw a rectangle over an item of clothing in an image, then get a match from a 20 million strong garment database with the opportunity to purchase from over 800 brands.

Put simply, there are tremendous opportunities in the here and now for brands to harness the power of image recognition to help them understand and connect with their target audience.

I’d advise you to get on it.


Jon Stubley is VP ANZ at GumGum.

Further reading

Feature image copyright: sergwsq / 123RF Stock Photo
Headhot image copyright: www.lumsdainephotography.com




You Might also Like

Leave a Comment