AI Makes Visual Intelligence Possible
AI Makes Visual Intelligence Possible
“The camera is already starting to replace the keyboard,” asserts Netra CEO, Richard Lee. The content that will dominate digital information flow will be visual, and for that reason image recognition is becoming a key component of marketing.
His company derives insight from visual data, fostering understanding of how consumers engage with brands through engagement with images. Netra is a leader in visual intelligence and search that uses machine learning to help marketers make sense of imagery on social media.
Some brands are already using image recognition to connect with and effectively market to their customers. They include Neiman Marcus. The upscale retailer offers its customers the Snap. Find. Shop app that enables them to use their phones to snap pictures of styles they like and find similar styles carried by the store. The app is demonstrated here:
The app allows customers to bypass typing in description of the clothing and rely on the image alone to convey what they seek. That kind of search is what we will be seeing more of in future, according to Lee. He predicts that we will see keyboards disappear, replaced by voice activation (as in the case of Amazon Alexa) or by something designed for a visual medium, “whether it's contact lenses or glasses.”
Netra software can detect about 4200 objects and scenes. With facial recognition, age, gender ethnicity, and other characteristics of people can be identified. AI can also be trained to detect brands and logos within a picture. This is a very important development as it frees such identification from attached text labels.
“All the analytics around social had been tied to text in the past,” Lee explained, offering the example of a brand hashtag like “#Coke” to point out the pictures in which the soda is featured. Now they no longer have to rely on such a hashtag to recognize that Coke is what person on the beach is drinking in the picture.
This “visual intelligence” can prove transformative, particularly because it makes it possible for brands to capitalize on what people seek, namely authenticity. As people tend to put greater faith in posts put up by friends than by brands, finding the brands in consumer pictures gives marketers a way to harness “microinfluence” with an impact that can be greater than that of a TV ads.
Today almost anyone can post good quality pictures and videos, giving marketers a treasure trove of content to mine. “You document your life through imagery and key life events,” Lee says.
Those images reflect not just what you're doing but “what's important to you.” When Lee uploaded his own pictures and ran analytics on the images, he realized that it really was “an accurate reflection of who I am.”
In his case, the pictures often center around two young children and their family activities. In this way, pictures capture a great deal of information about the customer's “activities, interests, demographics,” and even what they see as their “tribe,” that is the extended connections like “grandparents, friends,” or other important people in their lives, all of “which is really valuable.”
Netra has access to feeds that consumers opted into for public profiles or get access through partnerships with companies with feeds. On that basis, the company is working on “tagging a billion photos this month.” While that may sound like a huge number, Lee concedes that it really is “just a drop in the bucket,” in context of the 3 to 4 billion photos shared on a daily basis on social. All that processing is made possible by AI.
As to the question of real life applications, Lee says that Kantar, the second largest market research firm in the world, is “using us on the insight side.” In the past, it drew its data from consumer surveys and questionnaires, which had the drawback of limited feedback that would incorporate some bias. Now they “can tap into what consumers are organically sharing on social,” Lee said. That means that they not only can remove survey bias but also scale up the results, “extrapolating from billions t
Lee admits that in tackling image recognition, Netra is “competing with likes of Google, Facebook, Amazon, IBM, and Salesforce.” But he insists that theirs is “more accurate” and better at pointing out what businesses need to “extract value from it.” For example, for a picture of a table and six chairs, it could identify that what is represented is a business meeting. It can also identify a particular sports stadium in an image and even which teams are playing.
At the end of the day, it's not just about getting the picture but getting the insight of the story behind it. That's the value of visuals for marketing.