Internet Tech Blog

In searching for images using search engines like Google, Yahoo or other, the search results are often inaccurate. The reason for that may be because search engines can not index the images directly, which is the thing that indexed is only the image file name, alt tags (considered most important by Google, but many sites that do not include this tag) on image and text in a web page where the image is located.

Although there are now Tagging system for images, this system depends on how well each person in describing their picture with Tagging.

There are still many person who don’t know how to tags. And even there are many people who deliberately give the wrong tag, usually a keyword that is often searched so that the images are more often appear in search results. Another constraints are lazyness, how many images every day we upload to sites without tags.

To overcome the image search problem, product management director for Consumer Search Properties, R.J. Pittman give some information about the fututer search feature for images on Google.

In a video interview with Beet.TV, Pittman said that Google is making a visual crawling algorithm that can be used to search, identify and analyze the face of objects in the image file.

There also GeoTagging, a process of image location inserting, the possible start with digital GPS camera. Google already supports GeoTagging features with launched Panoramio site.

For more details, you can see on the video interview below:

Besides Google, there are also other sites that try to solve the search problem on the web. XCavator already use search system based on image composition.

How XCavator work:

1. Find a picture with the keyword as usual and click “GO”
2. From the results, drag one of the images that we think correct or accurate to the box above
3. Mark the part you want to search from the image you want to search, more signs that you provide, the more similar results with the image you select

View examples of how it works in the video below: