Proprietary Deep Fine-grained Recognition Technology

 

Magic? No, artificial intelligence and a brilliant team.

We've been busy researching and developing the most advanced fine grained image recognition based on deep learning technology and running on the world's biggest computers, specialized in picking up the most elaborate details from consumer goods and their shelves.

 
 

Uncovering the mystery

The Qube, our all-in-one artificial intelligence consists of many different engines working in harmony with each other to analyse shelf images instantaneously and extract information on product, brand, price and placement.

A look behind the cover

the different A.I. blocks created to perfection

 
qubesteps1.jpg

The Qube Process

Shelf images undergo a transformation through our multi-engine artificial intelligence

qubesteps2.jpeg

Advanced Image Stitching Engine

qubesteps3.jpeg

Image Reconstruction Engine

qubesteps4.jpeg

Realogram Engine

 

Anomaly Detection | Capturing differences and noncompliance between the ideal, the implemented and the reality

Novelty Detection | Capturing new scene entries like new products, changed packaging, misplaced random objects. 

 

qubesteps5.jpg

Deep fine-grained recognition engine

 

Feature Extraction | shapes, colors, text, borders, and other patterns to identify subtle details in products

Attentional Learning | Unsupervised machine learning to spot regions of interest

Contextual Deduction | Understanding the scene and deducing subtle differences in analysis (size comparison, package comparison)

section2_banner copy.jpg

Predictive Analytics Engine

 

Visuals and analytics data is aggregated and filed so that you can dive in and follow the indicators the matter to you:

On a dive-in SKU level and global overview

 
 

Predictive business intelligence

understand the impact of different merchandising levers on product performance

 
predictive-BI.jpeg