Face Analysis can Tell What You’ll Buy after Watching Ads

Image: Daniel Allan/Cultura/Alamy

System by Affective – Pick up Hidden Emotions

A software has been developed that can sense how much one tend to get swayed by marketing by analysing the face as one watches the advertisements. Ad companies seems interested in assessing the reactions of the consumers to their latest TV spots and this is done by bringing in few customers in the office and pose queries to them.

However, the system adopted by Affective, a start up in Waltham, Massachusetts, has the capabilities of picking up hidden emotions by monitoring face movements. The approach according to Daniel McDuff, Affectiva’s principal scientist enables you to find out what the consumer generally thinks from time to time as the ad tends to run and not what they say once it is ended.

He has commented that that `it provides a way of getting at those more genuine, spontaneous interactions and this is their instinctive response. It is not sent through a cognitive filter where they have to evaluate how they feel’.

The software of Affectiva tends to first identify important facial signs like the mouth eyebrows and the tip of the nose. The machine learning algorithms tends to watch how these areas move or how the skin texture and the colour alter in the course of the video. These alterations are then broken down into distinct expressions portraying shifting emotions.

Focus on Emotions – Surprise/Confusion/Happiness

McDuff together with his colleagues, in a study which had been published recently, has asked 1223 people to give his team access to their home webcams while they watched a series of ads for pet supplies, groceries and sweets.

Prior and after the running of the ads, the subject filled out on online surveys was on how likely they would purchase the products shown.As they watched, the software focused on the lookout for emotions like surprise, confusion and happiness.

The researchers then found that they could utilise the facial data to predict precisely, the survey results of someone indicating that they could depend on the analysis of the computer to know if an ad was successful.

McDuff is of the belief that in the future, the system could plug in TV services like Netflix. He states that one can imagine suggesting TV programmes or even movies which people could watch or ads which they would find enjoyable.

Subtle Disparities in Emotional Responses

The team at Affectiva had accumulated a database of more than three million videos of people of various ages, ethnicities and gender. McDuff is of the opinion that there could be subtle disparities in emotional responses.

For instance women seem to have more positive facial expressions than men. However on understanding the respond of various groups, companies could put together ads which could be fine-tuned for certain audiences. Similarly, the data could also be helpful to advertisers to pull their adverts to draw closely the emotions of the viewers.

Michel Wedel, studying consumer science at the University of Maryland in College Park states that automated emotion analysis systems seems to be promising enabling advertisers to cut down an ad, to find out what exactly works and what does not. He adds that it is particularly unobtrusive and does not depend on introspection or recollection.

To do the research through the viewer’s home webcams is an added advantage though it would not be fool-proof. People could be eating a sandwich or turning at the computer or turning their head and so unable to classify their emotions consistently.