We’ve created a two-part blog series about Artificial Intelligence & Digital Asset Management after testing the most advanced tools available. This post looks at the key features the major image-recognition tools provide and shows how they can be applied within Elvis.
For our top 3 things DAM users need to know before getting started, check out part 1 here.
We’ve been testing the functionality of the three key players in image-recognition AI – Google Vision, Amazon Image Rekognition and Clarifai.
Each tool has its own merits and provides different sources of data that can enhance the usability of your DAM. AI and Elvis were practically built for each other – not only were these integrations simple, they’ve also been incredibly effective very quickly. See it for yourself in the demo below.
The Key Features
This is the core functionality that Image Recognition AI provides DAM users. As you see in the video, we upload files within Elvis and the API automatically includes the metadata in the returned assets within Elvis.
At this point, it’s always important to validate the metadata – and build on it with additional information your users might need. From here users can either search, or filter by the tags you approve – providing multiple options to find exactly what they’re looking for. This could save thousands of hours (and a lot of frustration) for DAM administrators everywhere!
Clarifai has consistently delivered the richest, most varied metadata throughout testing, but the other tools are catching up quickly.
In our demo we’ve created specific fields for the different tools so you can see what each provides. You see already that there is a big difference between the type of metadata that is retrieved so when evaluating the different tools, it’s important to relate this back to your requirements check-list.
Imagine you’ve come across the perfect image you need online and you need to find something similar in your own database. Simply drag and drop that image to the plugin search function and Elvis will automatically deliver a bank of images with the same tags. The image you searched with will not even added to Elvis, meaning you keep a clean database of your own assets.
Here, we can filter by characteristics like mustaches to find the exact image we need.
If your image library features a lot of people, being able to break it down by searching for things like emotion, or personal characteristics can save significant amounts of time.
Amazon shines in this area, and with a simple integration within Elvis, it’s easy to keep this metadata as a separate field. By doing this it makes it very easy to refine searches by filtering, as you might with taxonomies.
Today’s AI software is powerful enough to identify major landmarks – and Google Vision will even append this with their exact co-ordinates. This benefits of this are twofold:
1. Users can search for the landmark by name without administrators having to manually add it
2. The asset can automatically be added to the Maps integration that already exists as standard within Elvis’ sample plugins.
In a similar way to the image-search feature, if you’ve already found a great image within Elvis but need more options, you can just right click and search similar by installing a simple plugin. With the extensive tags provided by AI this will help ensure users return the greatest volume of relevant results.
Image recognition and Elvis
These are just some of our favorite ways that AI may enhance DAM usability, but there will be plenty more to come as the technology evolves. Elvis is evolving just as quickly, and these integrations will be publicly available very soon.