We are at down of a new era:

Visual Inspection with Artificial Intelligence

How is your Retail Execution? 


Neural Networks

Artificial neural networks, a technique for implementing Machine Learning, were inspired by the model of a human’s brain that remains the most efficient and flexible (read: ideal) information processing device in the world. Neural networks technique that mimics the functions of neurons (understanding, cognition and perception) is currently considered as the best approach for solving computer vision tasks, speech recognition and generation, pattern recognition, music composition, medical image analysis, diagnoses and outcome predictions etc.

Machine Learning

Machine learning, a subfield of Artificial Intelligence, is a method of «training» an algorithm so that it can learn how to perform tasks that it has never solved before. The key principle of machine learning is: the more data the system has, the better it will be able to learn and function. The system is first «fed» with a massive amount of data and afterwards learns how to accomplish goals (e.g. distinguish between objects) and improve upon the process. Services developed by Kuznech are based specifically on the neural networks method.

Artificial Intelligence

Artificial Intelligence based on convolutional neural networks has modeled the human visual image processing part of brain (occipital cortex) in its propensity to pick out many points of interest of the object and create a «fingerprint of a target» — and not scan it from top to bottom or from one side to another. We use neural networks because they are useful to recognize visual patterns from pixel images and videos with minimal preprocessing and can recognize patterns with extreme variability and robustness to distortions.

We fully utilize the latest advances in modern GPU computing to cut rocessing times to a fraction of a second for images and a couple seconds for videos — a huge boost over the traditional client-server setup.

Merchandising Management​

Supermarket shelves might now be the world’s most expensive property. The more expensive the product representation on a shelf is, the more attention retailer/manufacturer pays to optimize costs for managing shelf space. Unfortunately, nowadays shelf audit is often carried out in manual way, which leads to numerous errors and massive profit losses.

We provide the service of automated SKU recognition and planogram compliance vs. realogram. Technology allows FMCG retailers to control SKU placement on the shelf, reducing manual work to an absolute minimum — sales representative only makes photos of shelves on a mobile device and then sends images to the server. The rest is done by Occicor.

Our service, being fully automated, significantly increases amount of data that can be collected by a sales representative, enhances analysis accuracy and notably reduces the time needed for processing and analyzing the results, allowing to evaluate your store performance at a glimpse.

The system analyzes images of shelves, recognizes SKUs and calculates KPIs, e.g.: On shelf SKU availability, number of facings, shelf-OOS (out-of-stock), compliance with the planogram, shelf share, competitive environment etc. — and generates an analytical report.

As a result, you get a highly qualitative information, speed up data processing, increase profitability and reduce transaction costs.

How it works...

Occicor automatically recognizes SKUs on the shelf, significantly increasing the accuracy and volume of merchandising data and reducing time and financial expenditures to process and analyze these results.

1) Take picture

Sales representative makes a picture/video of the shelf on a smartphone/tablet

2) Upload to cloud

The file is sent the server. SKUs on the shelf are automatically detected and recognized

3) Get results

Occicor creates a report with accurate and detailed information on: SKU presence, competitor’s products and other valuable metrics