Learning Visual Attributes

V. Ferrari, A. Zisserman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

Learning object categories has recently become a major focus of Computer Vision. In this work, we are interested in the complementary task of learning attributes, which are visual qualities of objects, such as red, striped, or spotted. To minimize the human effort needed to learn an attribute, we train models from web search engines, simply by querying them by the attribute name and passing the top few tens returned images to our learning algorithm (see positive training images on the right). Once a model is learnt, it is capable of recognizing the attribute and determine its spatial extent in novel images.
Original languageEnglish
Title of host publicationAdvances in Neural Information Processing Systems (NIPS)
Subtitle of host publicationVancouver, December 2007
Number of pages8
Publication statusPublished - 1 Dec 2007


Dive into the research topics of 'Learning Visual Attributes'. Together they form a unique fingerprint.

Cite this