Wouldn’t it be great if there was a universal language? A way to rapidly and effectively communicate with people anywhere in the world? Brands have been doing this for decades; religions and governments for centuries. The potential for images to instantaneously connect with groups of people is unparalleled, and fast becoming the default mode of communications online. On average, more than 2.5 trillion images were shared or stored online in 2016, of which 90% were taken by smartphones.
As camera and smartphone technology advance, expect image and photo sharing to become the default mode of communicating emotional and impactful content. Even now, who can forget the image of Neda Agha-Soltan, the Iranian woman whose death was recorded live online and shared as a series of memes to support the protestors during the Iranian Elections? Or the image of Syrian toddler Alan Kurdi washed up on a European beach? And especially the images of our most beloved brands?
Digital strategists are becoming increasingly aware of this trend and understand its importance to the evolution of the marketing and communications industries. While social media analysis is dominated by text-based analysis, the market for image analytics is expected to reach more than USD 30 billion by 2020. But why is it so important to analyse images?
Image analytics provide tremendous insight into brand health. Within the billions of photos we capture every day, many of them contain images of brands, influential people, products and logos. The traditional analytics process would have been to analyse the content of the text that someone used to describe the image or the ALT Tags embedded within. Now, with image recognition algorithms powered by Deep Learning, software is increasingly able to discern the content of the image and the context in which it resides. Such information, some of which is already used within our Facebook and Google accounts, is invaluable to companies for brand awareness, brand affinity, and ROI.
What’s In A Name?
As image analysis becomes more mainstream, it is important to know what it is and what it isn’t. Image analysis (the analysis of the content within a collection of images such as the surroundings, objects and more), is different to image recognition (finding images in an archive containing certain elements – normally logos). Image analysis can potentially analyse the vast online archive of images in a way that is useful to institutions. Want to discover and engage with new audiences that agree with your policies? Here’s a collection of images that features people happily doing what your policy focuses on. Want to figure out how to reach hard-to-reach communities? Search the online collection of images to get a cross-cultural, multi-national representation of how people feel about you.
However, the technology is not at the level it needs to be for organisations to make strategic decisions based on its conclusions. There are still issues in its ability to recognise the full spectrum of emotions – however, this is fast changing. To keep up with the change, we need to build new models for understanding visual social media, such as understanding the inherent value of images (such as selfies), and new models for perception. Images can be interpreted many different ways by two people, and this calls into question of whether the algorithms actually ‘see’ what is there? Or what they ‘want’ to see.
Ultimately, image analysis, as part of a social analytics toolbox, is already being used by organisations such as Snapchat, Facebook, Google, Samsung and more. It has the potential to provide deeper understandings of brand affinity, marketing effectiveness, public sentiment, and much more – all in a way that traverses cultural and linguistic barriers.
The way we analyse vast quantities of images will certainly be refined and integrated into current social media analysis offerings through the development of numerous APIs. However, the real value of image analysis will come when the combination of deep learning to provide real business intelligence from images.
What do you think?
Burns, A., 2015. “ Picturing the Social ”: Questions of method , ethics transparency in the analysis of social media photography . and. London School of Economics, p.1. Available at: http://blogs.lse.ac.uk/impactofsocialsciences/2015/06/10/picturing-the-social-method-ethics-social-media/ [Accessed April 28, 2017].
Cavalin, P., Figueiredo, F. & Pinhanez, C., 2016. Organizing Images from Social Media to Monitor Real-World Events. GitHub. Available at: https://flaviovdf.github.io/papers/cavalin2016-industry.pdf [Accessed April 28, 2017].
Garimella, V.R.K., Alfayad, A. & Weber, I., 2016. Social Media Image Analysis for Public Health. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems – CHI ’16. pp. 5543–5547. Available at: https://arxiv.org/pdf/1512.04476.pdf [Accessed April 28, 2017].
Comments on this article are closed.