This article is in continuation of my earlier blog post on the ‘Enhanced Smart Tags’, feature in AEM. In the previous post, I provided an overview of this feature and the requirements that it addresses. Here I discuss the recent improvements that have been made in the training process and the accompanying user experience with it . I’d strongly recommend reading my previous blog post before proceeding if not done already.
With recent advancements in computer aided image recognition technology, automatic image tagging systems have become ubiquitous. Adobe Experience Manager (AEM) has since long (AEM 6.3) provided the functionality of automatically tagging images with machine predicted tags aka Smart Tags. Smart Tags is powered by Adobe Sensei and features a tag prediction model, extensively trained on millions of Adobe Stock images. Beyond pre-trained generic image tagging, AEM also features a mechanism where users can create their own custom tag prediction models. Known as ‘Enhanced Smart Tags’, this feature is widely used for creating tag prediction models based on custom tagging requirements which fits the user’s business specific vocabulary.