ICDAR 2015 competition on text line detection in historical documents

This paper describes the ANDAR-TL competition (ANncestry Document Analysis and Recognition Text Lines) held at the 13th International Conference on Document Analysis and Recognition (ICDAR-2015). This competition uses the ANDAR-TL-1K dataset of 1300 images consisting of text-lines in paragraph form drawn from collections from the eighteenth and nineteenth century. These images are labeled with the coordinate location of the first character of the left-most word on each line (the origin point) and the objective of this competition is to use the training part of this dataset to develop and submit for evaluation a system that detects the locations of text-line origin points in the images from the validation set. Four teams submitted seven systems for scoring and evaluation. The results of this scoring are reported, along with brief descriptions of each system.