Feelimage.net - New Way to Search for Photos Online

Feelimage aims to help you find beautiful images on the web through both a word search and a "feeling" search. Believing that key word searches are not enough, Feelimage has created a new technology that is supposed to quantify colors like "red" and "blue" in images, as well as quantify time like "spring," "summer," "morning," etc. into images. Ideally, then, they want you to search for "red flower in the morning." I tried it, and well....it had mixed results. I got some flowers of all colors that seem to be morning or day shots, but it was impossible to tell the hour of day in the photos and certainly many of the photos were of yellow and purple floweres. So, I guess I am a tad skeptical of this technology. But it could be they are still working out bugs and at least this is a new approach to an old game ("find the photo!"). The public sites they are pulling photos from is Flickr, Photocorpus, Zorg and Photozou. The site was created and is still managed in Japan (you can ask for the site in Japanese).

"This is new image search site. The number of images on the internet is getting bigger and bigger these days. By the spread of digital cameras and cell phone cameras, our activity ' taking a picture' is much more common. Blogs and photo sharing services have been spread with the generalized broadband. It leads people to upload many pictures on the internet. Pictures used to be seen among only close friends, but now it's used as a communication tool for anyone. But finding pictures taken and uploaded every day is getting very difficult. If you can find pictures or images with your feeling, we believe you will find more interesting persons or things. Feelimage have been developed aiming a site you can 'feel' internet images."

I agree with the creators of this site that there are zillions of beautiful digital photographs online, many of which are in the creative commons for use by anyone. I also agree that using unique search terms and search technology is a positive evolution, even if it needs more time to fully develop.

I am still skeptical of their "feelimage" technology. I had mixed results, at best. Are they still doing research and development to improve this technology? Also, while they had good examples of tag searches for their "feeling" technology, I think they need to better explain how to refine a search using their preferences -- or at least highlight that option better on their home page. Altering the preferences would likely have big impacts on the results of the search, so I recommend emphasizing those options.