#research.microsoft.com

The dataset consists of 32.9 million distinct images, and each image is associated with one or more labels of the 18.7 million distinct textual queries. The correspondences between an image and a label (query) are determined by whether the image was searched and clicked by users under the query in a commercial search engine. Different from existing human-labeled datasets, Clickture covers most of the search interests of real-life users, in terms of both images and semantic labels. Moreover, click data is accumulating rapidly every day, thus it is feasible to extend Clickture with more images, labels and correspondences. Clickture can ...

Published by research.microsoft.com at 01/08/2013 7:00 AM EDT

The dataset consists of 32.9 million distinct images, and each image is associated with one or more labels of the 18.7 million distinct textual queries. The correspondences between an image and a label (query) are determined by whether the image was searched and clicked by users under the query in a commercial search engine. Different from existing human-labeled datasets, Clickture covers most of the search interests of real-life users, in terms of both images and semantic labels. Moreover, click data is accumulating rapidly every day, thus it is feasible to extend Clickture with more images, labels and correspondences. Clickture can ...

Published by research.microsoft.com at 01/08/2013 7:00 AM EDT

The dataset consists of 32.9 million distinct images, and each image is associated with one or more labels of the 18.7 million distinct textual queries. The correspondences between an image and a label (query) are determined by whether the image was searched and clicked by users under the query in a commercial search engine. Different from existing human-labeled datasets, Clickture covers most of the search interests of real-life users, in terms of both images and semantic labels. Moreover, click data is accumulating rapidly every day, thus it is feasible to extend Clickture with more images, labels and correspondences. Clickture can ...

Published by research.microsoft.com at 01/08/2013 7:00 AM EDT