The Power of Image Recognition in Digital Consumer Intelligence

Image Recognition applications spread from smart cars to medical apps, to support the visually impaired. In the field of Artificial Intelligence, it becomes increasingly central and also changes marketing and the market research world.

If you are already used to using images to make strategic decisions for your business, you are in the right place.
Watch our Digital Consumer Adventures webinars.

Experts from large companies, such as Deloitte, Group M and ENI have already found an innovative way to make the most of the data available from images for their market research.

It was 1990 when Tim Berners-Lee created the first browser, a fundamental step for building the Web as we know it now: in less than 30 years, that network that has generated enormous changes in the world which have been filled with words, conversations, post, comments… and of course, images.

If words have always been the subject of massive and in-depth analysis – from sociolinguistics to conversational analysis, through the semiotics of texts – the images are a little less so. Although visual communication has always been fundamental for the human being, images are a more complex and polysemic material, difficult to interpret.

Image Recognition is also changing the Web Consumer Intelligence: the possibility to analyze images through software will open several new ways for the marketers’ work in the coming years.

The amount of images shared on the Web is enormous: after all, the same Networks often entice users to upload images. The posts with images use to have a greater appeal than those ones made only by text. According to Hubspot, for example, tweets with images obtained 150 times more shares than textual tweets. An enormous value, if we think about the important informations contained in each User Generated Content, useful for a brand to understand purchasing habits and moments related to specific products.

No human has the ability to analyze billions of images online and extract insights from them. So, what exactly changes with the use of this technology?

Those who carry out Intelligence jobs often deal with millions of interactions and daily comments, solicited or not. The development of Image Recognition applications can help to skim the amount of data, eliminating spam, identifying the most important interactions and helping marketers to focus on the most significant data.

The images that users associate with brands can be very significant. Furthermore, elements such as trolls, distortions of the logo itself, offensive content, massive doses of spam can create the conditions for a crisis: it is so essential to be able to analyze not only the words but also the visual part associated with these messages, in order to prevent, as far as possible, the deterioration of relations with our public.

Image recognition technology can be very helpful. In what way? Screening millions of users, analyzing how they use the platforms and finding those ones who best fit the values of the Brand.

IR combined with chatbot technology can greatly simplify the work of Customer Care, automating procedures on ordinary problems and identifying the eventualities in which human intervention is required.

Image recognition linked to georeferenced data can be fundamental for all those activities that make geomarketing an important asset: from shops to tourism operators, through hotels and holiday sites, up to the organization of events.

A strong push to the study of images in contemporary times comes from the use of Artificial Intelligence: there are already several applications in the medical field, in the sector of intelligent cars or in support of the blind. A sector that has seen an unprecedented development thanks to the evolution of Machine Learning and Deep Learning.

Thanks to the growing importance of images, platforms such as Instagram and Snapchat have grown rapidly in recent years. In the meanwhile, IR has undergone a significant boost thanks to projects such as ImageNet and Pascal VOC: free image databases with millions of images associated with keywords.

The ability to analyze and recognize images represents a huge advantage for those who work on networks based exclusively on images such as Instagram: we can think about the number of daily selfies, the amount of visual content posted by foodies, the way of interacting with users who are passionate about fashion or design. At the same time, it’s possible avoid the dispersion of important messages that perhaps do not have any hashtag.

This is how MetaEyes works: it starts by selecting a hashtag and doing an exploratory search. Based on what the machine learns about visual content with relevant hashtags, it is able to find similar but free of hashtag images. This allows you to reach users who have not used any tags or keywords, by opening a channel that was previously unreachable.

Image Recognition can give a large number of opportunities to reach new targets, to optimize the interaction procedures with users, but also to streamline the daily work of analysts and marketers, by carrying out screenings and allowing them to work on most significant data.

Beyond the technological developments compared to the early studies of David Lowe (Credit: SIFT), the method of using this type of content is the thing that has really changed. From a long processing to a few clicks in a research environment on a SaaS tool, making these technologies available to anyone at any time, without any type of specific hardware.

Thanks to the Image Recognition technology and the automatic learning made available by SaaS tools, it is possible to identify patterns within any type of visual content and find the right answers to everyday questions.

If you are already used to using images to make strategic decisions for your business, you are in the right place.
Watch our Digital Consumer Adventures webinars.

Experts from large companies, such as Deloitte, Group M and ENI have already found an innovative way to make the most of the data available from images for their market research.