Computer Vision for shelf revision


Product detection and recognition in supermarkets

Client is a multinational company operating shelf revision in supermarkets. They requested our help to automatize product detection and recognition on their shelves.

The main features of the tool developed are:

  • TensorBox framework with inception v4 model for regression and classification problems.
  • OpenCV for visualization of the results, training set labelling tasks.
  • Deployment to the Google ML Engine for distributed training of the model.
  • Cascade recognition model to be able to recognize dozens of brands and subbrands among products.

The framework for object recognition is able to work on Google Cloud. Labelling framework for multiple purposes was built using OpenCV. It allows for a full cycle of creation, verification and testing of the training set of labelled images, each including about 200 objects.

This tool helps to recognize goods on supermarket shelves. By “digitizing the shelf,” the client is able to get role-based insights on a huge array of retail metrics that tell them exactly what’s happening on-shelf and what to do to ensure the best shopping experience and drive better sales.

TECHNOLOGY & TOOLS: AmazonOpenCVPythonSciPyUbuntuGoogle ML EngineGoogle StorageTensorfowCythonTensorBoxComputer Vision