Manipulating emotion with computer vision

Our recent blog post looked at Affective Computing in healthcare – a sector which has seen a wealth of research around emotional AI. Other areas are also investing heavily in the science of emotion to gain traction in ever-more competitive marketplaces.
Affective computing was pioneered by Rosalind Picard, an American engineer behind the companies Affectiva and Empatica. She published her first book on the subject in 1997 and has been a driving force behind the application of affective computing ever since. While the concept, therefore, isn’t new, the volume of research compared to the number of real-life applications prove just how complex the technology is. We take a look here at how emotional AI is being used to make marketing more targeted and merchandise more appealing.
Influencing your audience
How do we feel about a particular advert or campaign? Now the marketing gurus can tell us before we’ve even decided ourselves. Realeyes, a start-up created at Oxford University, uses webcams to measure consumers’ reactions to video clips. Sentiment and attention levels are captured and data is used to test for optimum content in adverts, identify a target audience and even to see which political campaigns are the most successful. Media partners for Coca-Cola, Heineken and Mars have experimented with Realeyes’ platform to hone their advertising and even select the best soundtrack.
A couple of years ago, Disney teamed up with Caltech to analyze audiences’ reactions to their movies. High-definition infrared cameras were used to record viewers’ faces watching blockbuster movies. Then, Deep Learning algorithms called factorized variational autoencoders, or FVAEs, were employed to translate these images into useable numerical data to understand and even predict an audience’s response to a particular scene. Word has it that this technology is being utilized to make Disney’s funny movies even funnier and ensuring that the sad parts really do tug at our heart strings.
What to watch out for
While still in its infancy, affective computing, empowered by computer vision and AI, is influencing increasingly diverse applications. Watch out for affective surveillance as computers learn to scrutinize large crowds for flash points, social robots trained to decrease tension and support mental well-being, and even menus that know what you want to order faster than you do as your eyes scan the content!
Want to know more about computer vision and AI? Sign up to our newsletter for highlights or contact our experts. Active Silicon’s range of camera solutions and advanced software can capture, process and manipulate images and video at high speed and high resolution, making them ideal for a wide range of applications.