Image Classification and Retrieval from User-Supplied Tags

This paper proposes direct learning of image classifica-tion from user-supplied tags, without filtering. Each tag is supplied by the user who shared the image online. Enor-mous numbers of these tags are freely available online, and they give insight about the image categories important to users and to image classification. Our approach is comple-mentary to the conventional approach of manual annota-tion, which is extremely costly. We analyze of the Flickr 100 Million Image dataset, making several useful observations about the statistics of these tags. We introduce a large-scale robust classification algorithm, in order to handle the in-herent noise in these tags, and a calibration procedure to better predict objective annotations. We show that freely available, user-supplied tags can obtain similar or superior results to large databases of costly manual annotations. 1