Portrait of a Technology Artist: Smart Cropping
By Ultan O'Broin-Oracle on Jul 15, 2014
By Joe Goldberg, Ph.D., CPE, Chief Research Scientist, Oracle Applications User Experience
Sometimes it’s the little things that make or break a user experience.
Consider a common scenario: You’ve just found the perfect image of yourself and uploaded it to your employee profile page. After a page refresh, your headshot is cropped to fit the vertical aspect ratio of its container frame. The only problem: The right side of your face is now missing because the technology cropped the upper left of the image. Unless you want to be mistaken for a Cyclops, you now must manually re-crop in Photoshop, or whatever’s handy, and upload again, hoping for better results.
Personal images are used widely in applications on the web, from social media profiles to human capital management portraits and public sector IDs. Note this tool from the U.S. Department of State’s passport application website.
What’s needed is a way to crop headshots so that faces remain fully intact.
Smart Cropping does just that. This Oracle Applications User Experience prototype feature automatically finds a person’s face in an image and then pads and crops the image to a desired aspect ratio so that the face is at the center of the cropped image. The technology that underlies Smart Cropping is OpenCV, an open-source computer vision library now being used extensively in consumer applications, such as cameras and cars.
OpenCV uses Haar Cascades from previously trained decision tree-based classifiers to rapidly find the coordinates of any faces in an image. This computer vision technology is transparent to the end user, who now enjoys perfect face cropping.
The image on the left shows a headshot that is horizontally centered, but vertically above the image’s center. The image on the right shows the output of the Smart Cropping feature, where the image is both horizontally and vertically centered.
OpenCV is a great example of how the energies of the open source development community, scientific UX insight, and consumerization of technology come together to meet user expectations about how things should work in an easy and familiar way to solve a common problem. This alignment of the social and scientific stars underwrites the Oracle Applications User Experience (OAUX) design principles.
Visit the Usable Apps website to find out more about how you can participate in shaping our UX and how to build great solutions using the same techniques and science that OUAX uses.
Interested in discovering more?
- About OpenCV
- About SimpleCV: An easy way to step into computer vision; wraps OpenCV and other libraries
- An excellent, practical book to get started using SimpleCV/OpenCV