akiwi is a semi-automatic image tagging system able to suggest keywords for uploaded images with minimal user input. akiwi was started as a student project by J. Hartmann, N. Hezel, M. Krause and A. Sonnenberg under the supervision of Prof. Kai Uwe Barthel.

akiwi uses visual search algorithms from pixolution to propose keywords using 22 million images from fotolia. Please observe that akiwi is a proof of concept without any commercial intent.

Have a look at www.visual-computing.com for other demos.

watch our demonstration video:

Impressum - Privacy Policy

akiwi is a tool trying to find keywords for your uploaded image.

In many cases akiwi will work almost fully automatic whereas in some cases it needs some help from you.

First click those images on the right that are similar to yours. Clicking once will confirm an image to be similar, clicking the same image again will deselect it.

If none of the images match, you should add a keyword that best describes your image. Click a keyword from the list or type a new one. You may also remove a keyword by clicking the x next to it.

Continue by selecting similar images and / or keywords until you are pleased with most of the keywords.

Finally click 'finalize' to refine your list of keywords before you may copy them to the clipboard.

Please observe that akiwi can only find keywords for typical stockphoto images. akiwi will not find keywords for celebrities or particular products.