作者: Timo Reuter , Symeon Papadopoulos , Vasilios Mezaris , Philipp Cimiano
关键词:
摘要: Nowadays, digital cameras are very popular among people and quite every mobile phone has a build-in camera. Social events have prominent role in people's life. Thus, take pictures of they part more them upload these to well-known online photo community sites like Flickr. The number uploaded is still proliferating there great interest automatizing the process event clustering so that incoming (picture) document can be assigned corresponding without need human interaction. These social defined as planned by people, attended for which multimedia also captured people. There an urgent develop algorithms capable grouping media depict or related to. In order train, test, evaluate such frameworks, we present dataset consists about 430,000 photos from Flickr together with underlying ground truth consisting 21,000 events. All accompanied their textual metadata. groupings been derived calendars on Web created collaboratively used Event Detection (SED) task was MediaEval Benchmark Multimedia Evaluation 2013. This required participants discover organize items event-specific clusters within collection documents. this paper describe how collected creation proposed evaluation methodology brief description challenge applied context task.