A Hanbury - Journal of Visual Languages and Computing, 2008 - dl.acm.org
In order to evaluate automated image annotation and object recognition algorithms, ground
truth in the form of a set of images correctly annotated with text describing each image is …