I am happy to announce that the following paper has been accepted to the
13th International Conference on Document Analysis and Recognition (ICDAR)
that will take place in Gammarth, Tunisia, on August 23 - 26, 2015:


               Using histogram representation and Earth Mover's Distance 
as an evaluation tool for text detection

      Stefania Calarasanu (1), Jonathan Fabrizio (1) and Séverine Dubuisson (2)

                (1) LRDE-EPITA, 14-16, rue Voltaire, F-94276, Le Kremlin
      Bicêtre, France
   (2) CNRS, UMR 7222, ISIR, F-75005, Paris, France

Abstract:

   In the context of text detection evaluation, it is essential to use 
protocols that are capable of describing both the quality and the 
quantity aspects of detection results. In this paper we propose 
a novel visual representation and evaluation tool that captures the 
whole nature of a detector by using histograms.  First, two histograms 
(coverage and accuracy) are generated to visualize the different 
characteristics of a detector. Secondly, we compare these two 
histograms to a so called optimal one to compute representative 
and comparable scores. To do so, we introduce the usage of the 
Earth Mover's Distance as a reliable evaluation tool to estimate recall 
and precision scores. Results obtained on the ICDAR 2013 dataset 
show that this method intuitively characterizes the accuracy of a 
text detector and gives at a glance various useful characteristics 
of the analyzed algorithm.



Ana Stefania Calarasanu
___________________________________________________
PhD Student
EPITA Research and Development Laboratory (LRDE) 

14-16 rue Voltaire, 94276 Le Kremlin-Bicêtre CEDEX, France
https://www.lrde.epita.fr/wiki/User:Calarasanu

_______________________________________________
Annonce mailing list
Annonce@lrde.epita.fr
https://lists.lrde.epita.fr/listinfo/annonce