The imageCLEF plant identification task 2013

Abstract

This paper presents a synthesis of ImageCLEF 2013 plant identification task, a system-oriented testbed dedicated to the evaluation of image-based plant identification technologies. With 12 participating groups coming from over 9 countries and 33 submitted runs, the 2013 campaign confirmed the increasing interest of the multimedia community in ecology-related challenges (respectively 10 and 11 groups crossed the finish line in 2011 and 2012). Contrary to the two previous years that were exclusively focused on leaf images, the coverage of the 2013 task was extended to six different types of view of the plant (flower, bark, fruit, entire, \dots) and significantly more plant species. This synthesis presents the resources and assessments of task, summarizes the retrieval approaches employed by the participating groups, and provides an analysis of the main evaluation results.