Div400

• DOI: 10.1145/2557642.2563670
• Publication Date: 2014-03-24T09:45:50Z
• Authors: Bogdan Ionescu; Anca-Livia Radu; María Menéndez; Henning Müller; Adrian Popescu; Babak Loni
• Keywords: Crowdsourcing; Relevance; Benchmarking; Modalities; Landmark; Relevance Feedback

In this paper we propose a new dataset, Div400, that was designed to support shared evaluation in different areas of social media photo retrieval, e.g., machine analysis (re-ranking, learning), human-based computation (crowdsourcing) or hybrid approaches (relevance feedback, machine-crowd integration). Div400 comes with associated relevance and diversity assessments performed by human annotators. 396 landmark locations are represented via 43,418 Flickr photos metadata, Wikipedia pages content descriptors for text visual modalities. To facilitate distribution, only Creative Commons included the dataset. The proposed dataset validated during 2013 Retrieving Diverse Social Images Task at MediaEval Benchmarking Initiative Multimedia Evaluation.