When you go onto google.com and search the images, you will get results that have your keywords in the text, but researchers at the University of California at San Diego are creating a special system in which pictures can be found according to what is actually in the picture instead of keywords that go along with it. As long as the system has been trained to recognize a certain image, it can be searched for and found under the contents of the picture itself. This software uses special algorithms to improve the image labeling. These algorithms are large part of the “Supervised Multiclass Labeling (SML) system” that is able to sort and tag each picture for the contents. Organizations could use this labeling and sorting system for image searches in order to better develop their websites. One main company that could really gain from this new technology would be Google.com, a search engine which currently bases the image search on keywords in websites, so the user gets a variety of different assorted pictures, many of which are not necessarily what the person wanted. This new labeling method uses a semantic image annotation that allows the contents to be sorted by image rather than text.
He main advantage to the new sorting system is that when a person searches for certain images, he or she will have better and more accurate results of the original keyword, since the search will not involve only the text involved with and attached to the image. This will make searching and finding the best picture for each need.
There really are no disadvantages to such a system since it is bringing higher accuracy to the image searches.
No comments:
Post a Comment