Advances in image recognition: recognising the power of an image
With over 1.8bn photos shared online every day we are living in the era of the visual web. From an insight generation perspective, the visual web provides a treasure chest of information into people and their lives with the photos people upload often revealing their passions and most joyful moments. They provide a literal window through which brands can get closer to and immerse themselves in the world of their audiences, often revealing brand changing truths.
As human beings we also process photos far more quickly and intuitively than text. Photos are thus incredibly powerful forms of insight communication as they tell the consumer story in a way which words or graphs alone cannot. Our brains are hardwired to relate to and connect with images more readily than with text; they convey sentiment and context in a way that no other form can and they provide a snapshot of what a consumer was doing, thinking and feeling at the time the photo was captured with no need for translation or explanation. Images by nature are also more memorable and shareable, particularly those that feature people.
The applications of insights from photos are also plentiful and varied. For example, an FMCG brand looking to launch a new breakfast snacking product can use the visual web as a powerful springboard for innovation. There are literally millions of photos online showing people from all corners of the world eating their breakfast. These photos tell interesting stories about what foods people are pairing together, where and on what occasions certain types of breakfast products are being consumed, and how different types of foods make people feel when they eat them. All of these insights can be derived without the need to ask a single question.
The photos people upload can also be incredibly useful in identifying trends e.g. what colours people are accessorizing with, and what patterns they are embracing etc. Under Armour used analysis of gym selfies, for example, to spot trends in gym wear helping to inspire the creation of a new range of clothing for its target audience.
When all of these unique attributes of images are added together they serve to amplify the value of photos and video to the insight process. And whilst marketers have long recognised the importance of creating striking visual content as part of their marketing strategy, it is arguably only more recently thatthe insight community has begun to harness the huge insight potential of the visual web.
With so much noise out there, a key barrier in the past has been that many brands simply haven’t had the bandwidth to be able to process all of this information beyond small scale qualitative analysis. It has also been said that many marketers have failed to address the shift towards the visual web in terms of analytics with many still analysing hashtags, text, comments and other tags to track what consumers are saying about a brand. Up to as many as 80 percent of images featuring brands are never surfaced or tracked according toAdWeek. Yet unlike explicit mentions via hashtag or text, photos offer more nuanced and potentially valuable insight into how a product is perceived by consumers. With analytics tools vastly out of sync with the visual web of today brand owners are therefore missing a trick.
Advances in image recognition technology are, however, changing this. Tools such as Mantii and Ditto Labs are offering exciting new insight possibilities through using artificial intelligence to quantify and codify the insights held within photos and images. They also offer various ‘cool tools’ through which to access these image derived insights such as visual analytics dashboards showing relevant photos by brand or product category which can be further assorted by theme, sentiment and location. Some of the tools also bring up information about the people behind the photo posts themselves e.g. their avatar and their social profile providing further richness and context. Marketers can also take this one step further and connect and engage further with the consumers identified as having the biggest visual influence on their brand.
2015 has seen significant advances in image recognition technology with big names like Microsoft, Facebook, Google and Apple all upping the ante in terms of artificial intelligence applications that mine this wealth of visual content. Artificial intelligence is getting more accurate by the day with brand logos able to be recognised even when partly obscured or upside down, for example.
When it comes to analysing the rich insights from photos (and video), although nothing can arguably replace the sophisticated nuanced interpretation of a human, from a counting perspective, image recognition technology has the power to provide the visual analytics bandwidth that up until now has proven elusive. Now that really is powerful news for the research community and for brands.