Total-Text (Total-Text)

Research Tasks

Text Segmentation in Natural Scene Images

2018-07-26 (v. 1)

Contact author

Chee Kheng Ch'ng

University of Malaya


This work is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported License.


The objective of this task is to label every pixel of a natural scene image to either 'text' or 'background' pixel. It is the same as the Task 2.2 of 'Focused Scene Text' challenge of ICDAR2013 Robust Reading Ccompetition[1].

[1] S. M. Lucas, A. Panaretos, L. Sosa, A. Tang, S. Wong, and R. Young. ICDAR 2003 robust reading competitions. ICDAR, 2003.



Similar to the aforementioned competition, both [2] and [3] evaluation protocols can be used to assess the performance of a scene text segmentation algorithm.

[2] Ntirogiannis, Konstantinos, Basilios Gatos, and Ioannis Pratikakis. "An objective evaluation methodology for document image binarization techniques." Document Analysis Systems, 2008. DAS'08. The Eighth IAPR International Workshop on. IEEE, 2008.

[3] Clavelli, Antonio, Dimosthenis Karatzas, and Josep Lladós. "A framework for the assessment of text extraction algorithms on complex colour images." Proceedings of the 9th IAPR International Workshop on Document Analysis Systems. ACM, 2010.


No comments on this dataset yet.

Add your comment

In order to comment on a dataset you need to be logged on
Register Now!


In order to rate this dataset you need to be logged on
Register Now!