Captioning photos using image recognition

I’ve been having a lot of fun with Google Goggles, an image search app for Android phones.

In my previous post, I included a few videos to quickly show off what it can do, but tonight I found a very practical use for the application. I’ve traveled to Europe a couple of times and taken plenty of pictures of famous landmarks, but I never got around to actually captioning them. I may have been able to recall where everything was immediately after the trip, but years later I don’t even know which country some of the photos (which aren’t geotagged) were taken in. Fortunately, one of the things Google Goggles is pretty good at recognizing is landmarks. Knowing this, I’ve been able to figure out where many of my pictures were taken simply by taking pictures of my pictures!

I recorded a video to show how this works. The landmark recognition bit begins about 4.5 minutes into the video.

10 thoughts on “Captioning photos using image recognition”

  1. Great use case!

    One note:
    Often if the audio book is the top result, the regular book will be listed in “other matches” farther down the page. Clicking on that will bring up the product search and books link as if it were the top result.

  2. Thx for the great tip! I was wondering, I imagine the image recognition is linked to a picture database? Do you know if this is linked to the pictures google users put up, or if it is something centralized done by Google? Just asking because I can imagine that image recognition is okay for famous cities, but not for tiny villages (in which I am living).
    thanks for any info you are willing to share on this.

  3. @Ignatia/Inge de Waard I’m not sure which sources of pictures are used. But even if images of your area aren’t searchable, if you’re at the physical location you want to know about, you may be able to use the nearby places overlay, which uses GPS.

Leave a Reply

Your email address will not be published. Required fields are marked *