Someday soon you’ll be able to log onto the Internet and search for images just by typing a description. Hold on you say, can’t we already do that. Well, yes, but the way the search engines algorithms work is by searching for the text that the image is tagged with. So for example, say you’re looking for an image of a man wearing with glasses wearing a white shirt and a hat. If you type that into Google the results you get will show what you’re looking for, but will also show you images that aren’t necessarily what you want. Someone could upload an image of a puppy and as long as they tagged it as a man with glasses, white shirt and a hat, you’ll see the puppy image included in your search results. But that is about to change thanks to image recognition.
Image recognition is the term used to refer to computers analyzing an image and determining what is in the image. So using our example above, when you search for the bespectacled man with the white shirt and hat the computer will search for images matching that description instead of the text tagged onto the image.
Searching for video will also become more accurate using this method. This works because a video is essentially many slightly different still images played at a high enough speed to give the impression of movement.
Of the many applications of image recognition, one of the most exciting may be product recognition. Imagine seeing an article of clothing or other physical product that you like. Now imagine taking a picture of that item with your phone and using an app to search for information on the product. This is what the Slyce visual search engine does. Using the app you can take a photo or use an existing product image to search for more information about the item.
We live in a time inundated with media and as time progresses we will be even more so. Image recognition will be the technology to help us sort through our digital lives.