Two hundred shades of white, or the application of image analysis in business

Did you kow that the Innuit have over 200 words denoting snow? This means that, as a result of living in utter whiteness, they have learnt to discriminate between its different features, including the incredible number of shades.

Their eyesight, considering the need to react to even subtle changes in the surrounding, has developed unique analytical skills which are not available to an average Polish, Spanish, Malaysian or the Republic of South African resident. What does it have to do with business?

Companies employing sales representatives collect masses of data and generate tons of reports, a lot of which are concerned with store shelf analysis – the number of products, their position, the so-called facing, shelf-sharing with competitors and many other indicators. All products have to be correctly recognized, distinguished and, obviously, described as precisely as possible. How to do this not only well but also fast? Technology comes in handy.

The technological hero of recent times, image recognition, has found its mission in business. The algorithms of image recognition have been harnessed to serve the  trend of marketing automation, automating the tedious activity of reporting on shelf stock levels. How does it work in practice? A photo taken with a smartphone is enough to generate a report being the basis for a decision to order goods, launch a special offer or implement some corrections.

OK, but how effective does this technology prove when confronted with the harsh reality of an untidy store shelf?

Round bottles, unevenly arrayed products
Arranging products on the shelf as if they were all posing for an IKEA catalog is the ideal of every manager that creating planograms. However, ideals do not exist. No later than an hour after aligning goods on the shelf and the visit of first customers, the “shelf cosmos” turns into chaos: bottles are turned around, cartons are askew and among ketchup stands, shaken (though not stirred :)), a small Martini bottle. The smart algorithms of image recognition come to rescue here, as they perfectly handle distortions and labels that are visible only fragmentarily. Therefore, you do not need to realign the products so they are recognized by the system. It is enough to take a photo with your phone and let the system do its job. Cool? I guess so.

Raspberry? Strawberry? Blackberry?
If you take a closer look at the labels of a series of products, you will surely notice they were created according to the same pattern, modified only in places (e.g. on juices we can see only changes of the fruit and name, the rest of the packaging remaining the same). It may take a while before you make sure that the juice you are looking at is actually tomato and not e.g. carrot juice. Technology also has to be able to differentiate between such similar fruits as strawberry, raspberry and blackberry. And so it is – and with 98% precision.

Kilometers of shelves…
Store shelves (only those to be reported on by one representative) within hours, days, weeks of work merge to form multi-kilometer motorways with roadsides of densely laid out products…All of them to be reported on. Assisted by the technology for image recognition and picture analysis, which examines the photographed products, we go along this road at a speed of 4 km/h. It is comparable to a walk through the woods… and this is exactly what managers should feel like after abandoning manual reporting in favor of a solution that is 15 times as fast. What is more, the application does not get tired – it is just as effective at the end as it was at the start of the working day (unless the battery in your mobile runs out, of course).

Professional photography for everyone
“Okay, but will the camera in my mobile cope with it?” Unless you have borrowed it from your great-grandmother, it will. The application assists the user in proper cropping and levelling, suggests changes of the angle and does well in low light. Therefore, additional stands or lighting are not necessary. The user is not bothered by grasping such notions as aperture, shutter or depth of focus. It is enough to find the shutter button.

Savings you can calculate
A sample calculation shows that for every 100 representatives a full examination of a store shelf ensures savings exceeding 460 hours monthly, which equals almost 20 working days. Is this a lot? To us, this seems quite a lot. Replacing manual with automated reporting would quickly save time and money for a trip of a lifetime, for instance to meet the Inuit in snowy Greenland.

The technology you are reading about now is called eLeader Mobile Shelf Recognition. This is a product created as part of the family of systems SFA / FFM / RSE eLeader Mobile Visit. If you would like to know how much you could save by using eLeader Mobile Shelf Recognition, use our calculator or order a demo presentation.