top of page

Smile, advertisers are watching: How marketing firms are scanning your social media photos for clues


In the era of social networking sites, it has become common practice to tailor our photos so we come across in a certain way.

But now there's a whole new audience that 'selfie' fans need to take into consideration.

Big brands have started using state-of-the-art computer software to scan the swathes of photos posted on sites such as Instagram for clues about their prospective customers.

Companies such as Coca-Cola are using a new image recognition software to track their logos - as well as 2,500 other 'details' - in the 1.8billion photos that flood social networking sites each day.

The software even gives brands the power to see if the customer is happy with the product - by reading and analysing the facial expressions of those in the photos.

The program, created by Ditto Labs in Massachusetts, also 'reads' the background, the person's clothing and even the location of the photo in a bid to glean as much information as possible as the customer and how they view the product or brand.

The wealth of information is then used to set up a profile which spells out exactly how that customer should be targeted by advertisers.

The pioneering software is a dream for marketing firms, allowing them to carry out extensive research while barely lifting a finger.

But liberty rights campaigners have warned the process goes 'far beyond' what most users should expect and that companies should seek permission before passing on the information to third parties.

Emma Carr, director of Big Brother Watch, a campaign group set up to challenge policies which it believes threatens privacy, told MailOnline: 'Social media companies are well within their rights to scan photos for marketing purposes, which will certainly come as a surprise to many users.

'We are all very used to receiving targeted adverts based on the information we provide about ourselves online, whether that be via our emails or status updates, but scanning our photos for logos and certain backdrops will go far beyond what many would expect companies to do with the photos we post.

'If companies want to use our data in this way, explicit permission should be sought. It is also only right that users ask for complete transparency about what data will be collected, analysed and who it will be sold on to.'

The program - which computer scientists began creating more than a decade ago - is already being used by brands owned by Procter & Gamble, which handles hair care company Pantene and Oral-B.

Other brands include Adidas, which, through the program, discovered that 13 per cent of its 'Adidas population' are also interested in Justin Bieber.

Ditto - which also has companies such as Colgate and Cadbury's within its portfolio - has also already worked with Budweiser, helping it find that beer drinking generally peaks at 11pm.

David Rose, chief executive, told The Times that the primary use of the software was 'listening', adding: 'So they know how the product packaging should be changed or who they should be aligning with for sponsorship.

'Say that you're Manchester United. How does Man U make a good case for bringing in new sponsorship? They can look at photos of people in Man U branded clothing and see what other brands appear.'

According to the site, the software 'unlocks brand data hidden inside the millions of photos uploaded each day'.

It claims to identify brands in every photo to 'make those photos actionable for marketers'.

The website explains that the company's 'proprietary computer vision' is trained to scan for 2,500 details in each photo, ranging from brand logos to fabric patterns.

The software can also detect where the photo is taken by using geocoding information. The geocoding uses longitude and latitude statistics to locate the exact position of the camera.

The company describes how the software can also 'access information about who posts images with your logo' by tracking 'faces in the image and their corresponding facial mood score (FMS)'.

Earlier this year, a team of Google researchers developed an advanced image classification and detection program called GoogLeNet

The programme can locate and distinguish between a range of object sizes within a single image. It can also determine an object within, or on top of, an object, within the photo.

Google has made the software open to other developers, to help increase its accuracy. In the future, the technology could be used to improve Google Image searches.

It could also scour YouTube videos for specific objects or shapes.

Featured Posts
Recent Posts
Archive
Search By Tags
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square
bottom of page