Retail Product Labelling Automation


Since 1982, The Warehouse has grown to be New Zealand's largest general merchandise retailer. The Warehouse is a Kiwi household name, with over 90 iconic 'big red sheds' around the country. They are part of The Warehouse Group, which manages market-leading retail brands including Warehouse Stationery, Noel Leeming, Torpedo7, 1-day and most recently TheMarket.

project SUMMARY

UVLens is the number 1 rated sun safety app on Android and iOS. UVLens shows the danger level of ultraviolet (UV) rays throughout the day and provides personalised recommendations based on the user’s skin type.

- 500k+ worldwide users
- Featured by Apple App Store
- Editors Choice on Google Play Store‍

The Warehouse manages one the largest retail inventories in New Zealand. Correctly labelling products with metadata is important as it enables personalised customer experiences and improves product recommendations on The Warehouse’s eCommerce store. However, the quality of data from suppliers can vary drastically, resulting in many missing labels and metadata for the categorisation and sorting of products.

Key challenges presented to us:

The retailer needed to keep track of DVD classifications, particularly whether DVDs had been correctly labelled or were missing the necessary age classifications. This tedious process was done manually, which involves an individual going through thousands of images and checking each DVD. The accuracy of this metadata is especially imperative, due to the prohibitions related to the distribution of age restricted media to underaged customers.

Another challenge was categorising toy products by age appropriateness. Many parents may not know what toys are appropriate for their kids, thereby relying on recommendations from retailers. The problem is most manufacturers do not specify age data, and is too costly for a staff to do this completely manually, which often involves a lot of guesswork.

The Solution

The data available was in an unstructured format, which is traditionally difficult for machines to process. There was an opportunity to leverage recent AI developments in order to automate the labelling process. 

For the DVD classification task, we made use of Computer Vision to detect product labels and automatically classify the ratings of DVD titles. This makes checking a database of thousands of DVDs a matter of minutes in the background. 

To sort the toys, we used Natural Language Processing to understand the description and predict the age range based on previously labelled data. This provides a scalable way for The Warehouse to continuously sort new toys into relevant categories.

The Outcome

project Details


The Warehouse



Want to stay in the loop? Subscribe to get updates from us.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.