Digital asset management (DAM) one of the most critical business processes for organizations today. Digitized content is everywhere, with one study predicting that by 2020, there would be 44 trillion gigabytes of digital content.
Thanks to AI-powered DAM solutions, many companies are already harnessing the power of computer vision to tag and organize their digital visual content. Picturepark, for example, used our technology to build the Clarifai Connector for Picturepark. This allows users to “seamlessly integrate artificial intelligence and auto-tagging” into the company’s content management platform.
Whether you are a DAM vendor like Picturepark, choose to go with a DAM vendor or decide to keep the process internal is up, you will still have to make a decision: do we build or buy? While keen research is necessary for you to determine which choice is right for you, a vendor-supported cloud API can allow businesses to create DAM systems for themselves and others, powered by computer vision.
So, why use a Cloud API for DAM?
Autotagging made easy and accurate.
With a cloud API, you can classify and tag droves of visual content in seconds, freeing up time and resources for tasks that are more remunerative for your business. For instance, Pacific Magazines is using computer vision to organize their archive of magazine photos. With Clarifai, they have so far tagged and organized over 10 million images, saving countless hours that would otherwise have been spent on manually reviewing each image.
With the Clarifai API, the company was able to accurately and comprehensively classify these images and can now take advantage of their extensive archives, analyzing their database of images to gain a perspective on cultural trends over the years. This information can then be used to potentially predict future ones (i.e., data mining.)
Search by keyword, visual similarity and beyond.
With thousands of images to look through, sometimes keyword searches aren’t enough. The best computer vision APIs allow for various types of searches. Clarifai Search, for instance, lets users search by keyword or concept, image, metadata, even geolocation, all within the existing API platform.
While APIs let you integrate visual search features on your site for your customers, they also allow you to utilize that feature internally by simply logging into your API account. For instance, when a content marketing writer at Style Me Pretty decides to write a blog post about “the most romantic wedding dresses,” such a broad concept will bring up a dizzying array of wedding looks. With visual search, the author can pick one or two inspirational images, and use that to sift through every wedding dress photo in their database quickly. They can even go a step further, combining visual search with the concept “romantic,” so their search is limited to dress photos that are not only visually akin to their starting image but also within the desired image category. With the relevant pictures in hand, the author can focus on writing the content their audience values.
Custom concepts? No problem for custom training.
Off-the-shelf solutions like Picturepark are highly valuable where your tags are multidisciplinary. Clarifai’s General Model, the basis of their Classifier, can recognize over 11,000 concepts across a wide range of subjects. However, these concepts may be too broad for individual needs. Architizer, for instance, needed a model that not only recognized construction or architectural-features like “cantilever” and “living wall,” but could name them. Similarly, West Elm's custom model uses its own terminology as does OpenTable.
While our team of data strategists was able to use their data and build the models for these companies in a matter of weeks, with a computer vision API, anyone can build and train a custom model themselves. From Waldo to Christmas sweaters, developers have used our API to build models for even the most specific of concepts. For the most part, computer vision can be trained to see and recognize anything humans can. Any company or industry jargon can, therefore, be utilized to generate such tags for your visual content.
Generate multilingual tags and metadata, with no translation required.
While custom training is one of the most valuable features of computer vision cloud APIs, sometimes the issue is language. For Staples Europe, for instance, their tags were generic, but they needed to cater to the diverse languages within their target market. By using Clarifai’s multi-language API, the tags and terms in their metadata were not just transposed into a different language, but actually auto-generated in the desired languages, capturing the nuances of each. Using the API allowed them to tag their entire inventory in each desired language in half a day, and saved them thousands of dollars in agency fees for translation services.
Connect to the cloud for data storage and consolidated learning
Cloud APIs not only allow you to utilize the vendor’s computer vision services, they also offer you access to world-leading cloud storage facilities. By storing your visual data on your vendor’s remote servers, you won’t need to worry about the activities associated with maintaining hosting facilities, including data protection, and so can focus on tasks that actually generate revenue.
Further, cloud APIs consolidate the knowledge gained by all the computer vision models stored in the cloud, to which every model can subscribe. With this feature, your DAM solution can learn from other models in the cloud, making for a smarter application without any intervention from you.
Whether you choose an existing DAM platform or go about the process yourself, computer vision has already shown itself to be of great use for managing data assets. With a cloud API, however, you can build a customized solution for your business,