Designers have a new tool to add to their bag of tricks.
Image recognition means that computers gain information and understanding from digital images or video and they also analyze and categorize image data. Designers can now search their files by image instead of just text and automatically add metadata tags to images in their libraries — for example, a hideaway door on a kitchen cabinet. A meta tag or metadata is information that is not shown in the image but is attached to the file such as descriptions, keywords, document authors, last modified dates, etc.
Image recognition, a type of artificial intelligence, is more advanced than just identifying an image; it also has the capability to identify context. Instead of identifying “two people in a kitchen,” computers (like smartphones) can recognize that the two people are at the kitchen sink with plates in their hands and the water is turned on; thus understanding and conveying that people are doing dishes.
Below are some areas where designers can benefit from image recognition, which is just in its infancy.
Automated Image Tagging. You can have your image files tagged with meta-data automatically by a cloud service like Google Drive or Box so they are easily searchable. You can literally dump tons of photographs into one of these cloud services, and it will attach meta tags for you using artificial intelligence. I was stunned to look at my image files recently and see all the added meta tags. If you’ve ever been in a design meeting with clients, and you know you have the perfect image to share with them – but you can’t find it right then – image recognition will be a game changer for you.
Reverse Image Search. You can now search online by uploading an image to a service like google reverse image search. I have use this to find manufactures of wallpaper and ceramic tile in less than 30 seconds. I’ve also use the search to identify products such as Kohler sinks and metal finishes like copper for hoods.
Discovery. Designers love discovery and new inspiration, so for this example, I’m going to compare a search for a piece of cabinetry hardware by text and then by image. By text, I type “cabinetry hardware” and all my results down at the bottom of the page are cabinetry hardware (below).
Next, I’m going to upload a piece of hardware to start the image search, and although cabinetry hardware does come up, check out what else the computer found based on the image and the “nickel” meta tag .The screenshot of other nickel images is well down on the page, but on the way there, I see images and shapes that I’ve never thought of using for cabinetry hardware (below).
For the whole history of computing searching has been just by text, which is narrow because you’re searching for something very specific. Just the opposite is true with images because images can convey many things all at one time. Image search has open the door for discovery.