The Interesting Part of The Image In Focus
Responsive design supports a host of different device sizes, and the popular multi-channel approach further increases the number of differently cropped versions. Our teams rarely develop web applications with fewer than a dozen different image formats. Depending on the context, these applications will automatically select the portrait or landscape format with the right dimensions.
Due to the sheer number of required image formats, an automated, software-based calculation to help lighten the workload of image editors and to ensure consistency across the platform over the long term is often required. When automatically cropping images, it helps to use a focal point. This ensures that the interesting part of the image always remains in focus.
Automated Focal Point Calculation With AI
How can AI support content managers? Typically, the focal point is determined manually in the Content Management System. If large numbers of images need to be processed, for example in magazines or for user-generated content, it may make sense to have AI do an automated calculation. These services use machine learning algorithms to analyze every uploaded image and provide the necessary information to automatically determine the focal point. If necessary, the editors can then change the focal point. Apart from the time saved, there is another advantage to this solution. Setting the focal point automatically makes image editing easier for people with limited motor skills who have difficulty using a mouse.
Other Applications for AI in Image Management
AI can also help set keywords and descriptions for images. Images can be categorized automatically, which once again lightens the workload of the content manager. The images are easier to retrieve in the Content Management System, which creates new opportunities for linking them in the front end. A color scheme recognition function automatically sorts images by color.
The automated calculation of the focal point increases usability and accessibility for the editorial department. Thanks to AI, the editorial staff can now focus on what matters when processing images for an increasingly complex front end with various device sizes and channels, i.e., the content.
As part of the relaunch of MySwitzerland.com, we implemented the “Cloudinary” AI service for efficient image management in Sitecore 9.
Smart and Automated: Image Editing at Switzerland Tourism
Guest Contribution by Markus Dittli
Head of Digital Platform & Project Management
At MySwitzerland.com, we want visitors to start dreaming. More than 400,000 images and videos showcase the beauty of Switzerland.
The vast variety of devices and the high demands on performance are key drivers for the management and processing of assets: All images need to be available in various formats and sizes and must be optimized for different browsers, devices and connection speeds. The extensive linking of contents on MySwitzerland.com through different teaser elements is an additional challenge, and cropping the image correctly is not easy. In addition, we use a lot of images from various content partners alongside our own assets and display them on MySwitzerland.com.
This all means that managing images is no longer a task we can handle manually. To address this, we have integrated the Cloudinary AI cloud service. Cloudinary uses machine learning to analyze every image and creates the required derivatives based on various parameters. Cloudinary’s image recognition algorithm decides autonomously on the maximum possible compression and appropriate cropping. Editors only have to enter one image into Sitecore: The editing is carried out by Cloudinary. For us, this eases the burden considerably. We have more time on our hands to generate and link content and to keep up the quality. At the same time, we see an increase in the degrees of freedom for the design elements on the website, because image formats and sizes are adapted flexibly and automatically. And just in case the cropping ever goes wrong, editors can still crop images manually.
For us, this is a first venture into the realm of artificial intelligence – additional applications will be sure to follow!