Here are some random items I had sitting around that I thought would be interesting to search for using the Google Goggles application on my Nexus One.
I repeated a couple of the searches after the video was done and actually had some more fitting searches show up. What was the difference? The thing I realized was that you get far better search results when the entire item you want to search for is in the picture...for instance, if you are taking a picture of a book or a box, zoom in close enough to get the entire cover of the book so the edges of the item are not "scanned" as part of the image (the problem with including the edging is that you also then capture the background images/colors which only makes the image more complex and less specific to the item you want to search for). Another thing I observed was that complex items with a lot of texture tend to scan more effectively because there's more contrasting lines (which increase uniqueness) and makes it easier to clearly define what "it" is.
@adam_maikkula
Link to this comment:
All Comments (0)