AI image recognition to identify products within supermarkets

Client

A leading consumer goods company

Background

To better monitor product visibility in stores, our client wanted to implement an automated solution for capturing and analysing images of their goods on supermarket shelves.

The ideal solution required a custom image classification model capable of identifying the client's products in varied conditions. Our client selected Audacia to collaborate on a pilot leveraging Azure Custom Vision, assessing its suitability for production deployment.

Solution

Audacia worked closely with the client to build an image classification model using Azure Custom Vision.

Azure Custom Vision, a part of Azure Cognitive Services, was used for this project because it provides built-in functionality for identifying products on shelves. Azure Custom Vision provides granular functionality for choosing what machine learning you want to create, categorised into:

• Project Types: Allows users to select whether they want to classify images (Classification) or detect objects (Object Detection). The Object Detection Project Type was leveraged in this project to train the Custom Vision model to recognise individual products.
• Classification Types: Distinguishes between Multilabel or Multiclass. Multilabel means that any number of tags can be applied to the images, whereas multiclass organises images into single categories. This means that most of the images that are uploaded will be organised into the most likely tag.
• Domains: Particularly useful is the extensive list of domains that Azure provides. Users can select from this list to focus their solution around one area of computer vision. Azure already has a base model that has received a lot of data on how to identify images of, for example, food, landmarks and other areas.

Key steps included:

• Collecting a dataset of product images from varied angles and conditions
• Labelling images with tags for the model to learn from
• Training a detection model optimised for identifying consumer goods
• Testing and refining the model using an iterative feedback process
• Deploying the model via a web app for analysis and visibility

Built entirely in the cloud, the Custom Vision model analyses real-time images and accurately identifies the client's products on shelves. The pilot demonstrated the ease, speed, and flexibility of the Azure Cognitive Services platform.

Results

Custom Vision enabled the creation of a proof of concept by creating, training and refining an image detection model. The implementation of Azure Custom Vision provided the client with a clear proof of concept for how an AI solution for product image detection would behave. 

The Azure Custom Vision proof of concept delivered:

• Accurate classification of products from real-world images
• Rapid model development and training
• Easy refinement using an intuitive web interface
• Flexible deployment options for production
• Increased understanding of Microsoft's ML offerings

By leveraging Custom Vision's advanced machine learning capabilities, the client gained data-driven insights into product visibility and shelf placement.