Welcome to ICDAR2011 Robust Reading Competition
Document Image Analysis (DIA) was established to address the need for the automated analysis of scanned paper documents, but has since evolved to incorporate a great variety of text containers that include scene images, video sequences, and digital-born (Web and email) images. These types of text containers introduce a range of different problems for document analysis such as anti-aliasing and compression artefacts. Moreover due to their nature have forced a shift of paradigm in the Document Analysis field as we now have to deal with text in complex colour contexts as opposed to the traditional black and white documents.
The ICDAR 2011 Robust Reading competition will carry on from the tradition of previous similar competitions organized in ICDAR 2003 and 2005 but at the same time it will offer a much broader test-bed for participating authors. Two challenges will be organized focused on one hand on camera captured Scene Images and on the other hand on digitally created Web and Email Images.
While scene and born-digital text images appear on the surface very similar to each other at the same time they are distinctly different. Born-digital images are inherently low-resolution (made to be transmitted online and displayed on a screen) and text might suffer from compression artifacts and severe anti-aliasing. Scene text images on the other hand are high-resolution camera captured ones and present high illumination variability and perspective problems.
Therefore it is not necessarily true that methods developed for one domain would work in the other (or is it?). This year we have set out to find out through a the ICADR 2011 Robust Reading competition that contains challenges on both domains. We strongly encourage authors interested to participate to submit results to both challenges! This would allow us to do a meaningful comparison, while it involves minimal effort from the part of the participants (both challenges are structured the same way). Plus you get a second chance to win!
The challenge is set up around three tasks: Text Localization, Text Segmentation (Born-digital images only) and Word Recognition. Participation in any or all tasks and in one or both challenges is welcome! Check the individual challenge Web sites for information.
The goal of this competition is to compare and evaluate the state of the art in Text Recognition systems. The contest will consist of two primary challenges:
- Challenge 1: Reading Text in Born-Digital Images (Web and Email) http://www.cvc.uab.es/icdar2011competition/
- Challenge 2: Reading Text in Scene Images http://robustreading.opendfki.de/wiki/SceneText
Datasets and ground-truth information will be prepared such that participants can choose to take part in any of the tasks independently. The competition will be organized in open mode. The test data will be made available for 3 days on the respective challenge websites and the participants will be asked to upload the results of their algorithm for each of the sub tasks till the deadline.
- S. M. Lucas and et al, ICDAR 2003 Robust Reading Competitions: Entries, Results and Future Directions, International Journal on Document Analysis and Recognition, vol. 7, pp. 105-122, 2005.
- S. M. Lucas, ICDAR 2005 Text Locating Competition Results, in Proceedings of the 8th International Conference on Document Analysis and Recognition, Seoul, S. Korea, 2005, pp. 80-84.
- C. Wolf and J. M. Jolion, Object Count / Area Graphs for the Evaluation of Object Detection and Segmentation Algorithms, International Journal on Document Analysis and Recognition, vol. 8, no. 4, pp. 280-296, 2006.
German Research Centre for Artificial Intelligence DFKI Kaiserslautern, Germany
Computer Vision Centre, Barcelona, Spain