| Website | riya.com |
| Blog | blog.riya.com |
| Category | Web |
| Phone | 650-638-9705 |
| Employees | |
| Founded | 7/07 |
| Total | $19M |
| Series A BlueRun Ventures Leapfrog Ventures Peter Rip John Malloy | $4M |
| Series B, 1/06 Bay Partners Neal Dempsey | $15M |
Riya is a visual search engine. Users can use it to search their photos, their friends’ photos, and public photos on Riya.com.
Users must first sign up at Riya.com to download and install the Riya uploader. Once installed, users can upload their photos. As expected, users can control privacy settings of various images being uploaded. The client will auto-rotate photos with faces and text, resize large photos to 800 x 600 resolution, and detect faces within the photos.
In order to take full advantage of the facial recognition capabilities of Riya, users must first “train” Riya – there are three ways to train Riya to recognize faces. Users can drag and drop faces to tags or tags to faces; they can use Auto Training to inherit the training sets from their contacts; or, they can individually tag photos while browsing through their collection.
Riya searches users’ photo collections to auto-tag the faces that match those they have trained. It sorts the results into three categories: Definitely a Match, Maybe a Match and Not a Match. When Riya has gone through the entire collection, users will see the results of the auto-tagging. Riya is not perfect, so it’s bound to make mistakes from time to time. Users can easily click on the photos that are not a match to eliminate them from the set.
Users can browse by People, Albums, Time, Location and Text or perform a search for a specific term. There is no limit to the number of photos one can add to Riya visual search.
| Website | like.com |
| Blog | blog.like.com |
| Stage | Live |
| Tags | imagerecognition, visualshopping |
Unlike most other image search engines (including Google image search), like.com indexes images based on more than simple metadata and tags. Using Riya’s image recognition technology, like.com can search through a library of products and find ones similar to the look of any selected product. Users can shop for shoes, bags, watches, jewelry, clothes and much more without even typing in search terms if they choose.
For image search queries Like.com matches users’ searches with similar or identical objects by comparing each object’s “visual signature”. These signatures are “mathematical representations of images using 10,000 variables”.Currently there are two ways to search Like.com with at least one more planned for the future. Users can click on products displayed on the Like.com page to find similar products, or users can type in a descriptive phrase. Typing “Collar Long Sleeve Shirt” will prompt Like.com to translate those descriptive words into mathematical variables which constitute an object’s visual signature and then find similar signatures.
In the future, Like.com will allow users to upload images as search queries expanding the number of possible searches to any product image online.
