Vearsa – Product detection

Detecting book covers online


The ebook platform Vearsa aims to rethink the publisher’s sales opportunities by tracking book retailers’ websites. Hence, it needs to regularly check that all the preview images of books covers displayed throughout the internet are the accurate covers released. The client needs to optimise the resource dedicated to image checking for its database of more than 500 000 ebooks available.

Key performance indicators

+50K images in
Average of 4,5 million
of annual queries
+300% annual growth

How we help

Vearsa has deployed an automated system integrating the LTU’s visual recognition API.

First of all, thanks to a crawling tool, the images and metadata of the books detected in the product page of the targeted marketplaces are collected. Secondly, it operates a matching process using our image recognition API to identify among the collected book covers those that correspond to the reference images in Vearsa’s database. And thus detect the collected covers images that do not match with the images in the Vearsa’s database, indicating that the image displayed on the product page is not compliant.

The benefit for Vearsa is a considerable saving of time and manpower, thanks to an automated process capable of obtaining reliable results in a few seconds for several hundreds of thousands of processed images.