|
contents | software | |||||||
| Riya Photo Search Engine Goes Into Public Beta Riya, the first photo search engine, announced from the stage of the DEMO Conference that it will go into public beta this calendar quarter. Riya uses proprietary face and text recognition technology to let users search for photos of themselves, people and places athttp://www.riya.com. You can use the service in one of three ways. First, you can search the public collection for a wide variety of photos. Shah believes that as the peer-to-peer recommendation trend grows, more people will be using real photos by real people to research future purchases. The second way you can use Riya is to search your own collection after uploading your photos and training Riya to recognize people. Training takes a few minutes of identifying faces on a few of your photos before Riya looks inside of your entire collection to find the same people and automatically tag them. Riya's magic is in its proprietary face recognition technology that uses various contextual clues to enhance the results of the automatic tagging. For example, Riya uses the clothing of the subject, where the photo was taken and who else is in the photo. It also uses text recognition to read street signs, names on conference badges and any other words inside photos. Riya is designed for recognizing up to 400 unique people in personal digital photographs. It works best if photographs are high resolution and include time and date stamps, common to the output from most digital cameras. It also works best for identifying the people who appear most often in your collection. You can also search, share and get photos in your friends' collections on Riya. The social element of the search not only benefits you by adding another layer of privacy (you have the choice of making your photos viewable by the public, friends and contacts, or just yourself), it also helps you train Riya. By indicating that another member of Riya is a contact, you automatically inherit the any training your contact has done on his/her set. Any shared people you have in your collections will be automatically tagged in yours. write your comments about the article :: © 2006 Computing News :: home page |