Overview of the ImageCLEF 2012 Flickr Photo Annotation and Retrieval Task

ImageCLEF's Flickr Photo Annotation and Retrieval task aims to advance the state of the art in multimedia research by providing a chal- lenging benchmark for visual concept detection, annotation and retrieval in the context of a diverse collection of Flickr photos. The benchmark con- sisted of two separate but closely connected subtasks, where the objective of the first subtask was to accurately detect a wide range of semantic con- cepts for the purpose of automatic image annotation, while the objective of second subtask was to correctly retrieve relevant images for concept- oriented queries inspired by what people actually search for on the inter- net. This paper presents an overview of the benchmark, summarizes the annotation and retrieval techniques proposed by the participating teams and evaluates their performance.