We're fresh at Geary Interactive.
Subscribe to our blog to get a taste of our perspectives, ideas and endless ingenuity.
-
Image Search That Beats Google and Yahoo
Posted on April 4th, 2007 No comments
At the UCSD Jacobs School of Engineering, some electrical engineering students have come up with what looks to be a good start toward the future of image and potentially video search engines(Read More Here). Their image labeling and retrieval system automatically identified the above picture which it had never seen before with the labels: “Cat, tiger, plants, leaf, grass”. The major search engines currently rely on meta information (human generated labels and descriptions) to determine the contents of images and video. I think we all look forward to the future of image search when I can type a friends name into my electronic photo album, and automatically view all the pictures with that individual’s smile!
2 responses to “Image Search That Beats Google and Yahoo”
-
In related fashion, Google has started its “Image Labeler” allowing users to label random images to help improve the quality of Google’s image search results. What’s fun and interesting about it is that it pairs you up with another random user and over a 90 second period you try and match your labels to random images displayed. Google keeps a tally of daily winners and all time best labelers. Take a look!
http://images.google.com/imagelabeler/
-
Jeff Coleman April 4th, 2007 at 15:02
This is certainly a step in the right direction, but two areas will be quite challenging for cataloging images and video:
1) intent – What does the image portray? A tiger resting, in this example
2) perception – Is it a beautiful picture? Is it a tranquil scene?
Images and video will remain much more complicated than text for quite some time.
Leave a reply
-
Jason April 4th, 2007 at 15:02