The Digital Economy Development Institute (DEDI LLC), a resident of the Skolkovo Foundation’s IT Cluster, has released a new version of the SmartLook image recognition system that supports iPhone and iPad mobile devices. The system is developed on the basis of the SmartLook Engine digital platform and is designed to automate the collection of merchandising information and calculate KPIs for the quality of work with retail outlets.

The SmartLook Engine platform is used in the design of software for object recognition from photo or video using machine learning (artificial intelligence) technologies. The SmartLook Retail system was created on the platform, which solves the problems of recognizing and analyzing goods and price tags on store shelves and POS materials in retail outlets (light panels, banners, promotional stands, etc.).

Now, in addition to Android smartphones and tablets, the mobile neural network works on iPhones and iPads running iOS 12 and later.

The SmartLook Retail system has already proved its effectiveness by increasing labor productivity by at least 20% of merchandisers from JTI, Ferrero, Schwarzkopf and other world and Russian manufacturers of FMCG. Thus, more than 800 mobile employees of the Mars company work with the SkyNet Retail recognition system in 7 countries (Kazakhstan, Belarus, Kyrgyzstan, Uzbekistan, Armenia, Georgia, Azerbaijan). Implementation is planned for the next year in Moldova, Mongolia, Tajikistan, Turkmenistan and Turkey. Mars receives 5 times more marketing data with the system than before. Now the assessment of the representation of goods is carried out in the context of each commodity item (SKU) for own goods and products of competitors.

The SmartLook Retail new version recognizes objects on a mobile device in just 1.5 seconds, which, in comparison with well-known competing solutions, significantly saves the most valuable thing - the time of a mobile employee, which is more than 15 minutes per visit to a retail outlet. At the same time, the recognition accuracy reaches 98%.

The SmartLook Retail mobile application recognizes objects without the Internet and expensive servers and instantly provides the merchandiser with recommendations for laying out and replacing price tags and POS materials. Only the control of correction of the display and prices on the spot, without delaying time, creates the conditions for ensuring the maximum economic effect for both manufacturers and retailers. Currently, retail chains are testing the recognition system in order to automate internal merchandising processes.

The solution based on the SmartLook neural network works as part of the Optimum mobile SFA system or as an independent cross-platform mobile application. In addition, developers of third-party recognition systems can embed SmartLook into their applications.

Khuseyn Az-zari, DEDI’s General Director adds, “We are constantly improving methods and algorithms for recognizing objects in the image. Recent results in practice have shown the accuracy of identifying goods 98% and higher. In addition, we are adding new tools to the system, for example, the highly demanded calculation of PerfectStore indicators based on recognized photos with detailed reports down to SKUs”.