ICDAR2015Robust Reading Competition

Introduction

The results of the 2015 competition were announced during ICDAR 2015. You can read the final report and download the presentation here.

 

"Robust Reading" refers to the research area dealing with the interpretation of written communication in unconstrained settings. Typically Robust Reading is linked to the detection and recognition of textual information in scene images, but in the wider sense it refers to techniques and methodologies that have been developed specifically for text containers other than scanned paper documents, and include born-digital images and videos to mention a few.

Robust Reading is at the meeting point between camera based document analysis and scene interpretation, and serves as common ground between the document analysis community and the wider computer vision community.

The ICDAR Robust Reading Competition has been held five times [1-5], in 2003, 2005, 2011, 2013 and 2015. The competition is organized around challenges that represent specific application domains for robust reading. Challenges are selected to cover a wide range of real-world situations. Each challenge is set up around different tasks.

The ICDAR 2015 Robust Reading competition builds upon the success of the previous editions. The main changes that were introduced for the 2015 editions are summarized in the table below.

OverviewTable_v03.png

For the 2015 edition we have introduced some major changes.

First, a new Challenge on Incidental Scene Text (Challenge 4) is introduced, based on a dataset of >1500 images acquired with wearable devices. Incidental scene text refers to text that appears in the scene without the user having taken any specific prior action to cause its appearance or improve its positioning or quality in the frame. While focused scene text (see Challenge 2) is the expected input for applications such as translation on demand, incidental scene text covers another wide range of applications linked to wearable cameras or massive urban captures where the capture is difficult or undesirable to control. You can read the task details about the new Challenge 4 here.

Second, we have introduced an "end-to-end" task in all competition Challenges. The new end-to-end task will aim to the simultaneous localisation and recognision of all words in the image. More information about this task is provided in the respective task pages:
Challenge 1Challenge 2 | Challenge 3Challenge 4

Third, we have introduced new video data for Challenge 3. See the details here.

The competition for 2015 will focus on the above new or updated tasks (red in the table above). Therefore the active tasks for 2015 are:

T1.4, T2.4, T3.1, T3.4, T4.1, T4.3, T4.4

Users are welcome and encouraged to participate in older tasks which remain open for submitting results in a continuous mode. Old tasks will remain availble for the community but will not be analysed for the ICDAR 2015 report and no new awards will be given out during ICDAR 2015. These are tasks: T1.1, T1.2, T1.3, T2.1, T2.2, T2.3

  1. D. Karatzas, L. Gomez-Bigorda, A. Nicolaou, S. Ghosh, A. Bagdanov, M. Iwamura, J. Matas, L. Neumann, V. Ramaseshan Chandrasekhar, S. Lu, F. Shafait, S. Uchida, E. Valveny, "ICDAR 2015 Competition on Robust Reading", In Proc. 13th International Conference on Document Analysis and Recognition (ICDAR 2015), IEEE, 2015, pp. 1156-1160. [pdf] [presentation]

  2. S.M. Lucas, A. Panaretos, L. Sosa, A. Tang, S. Wong, and R. Young, "ICDAR 2003 robust reading competitions", In Proc. 7th International Conference on Document Analysis and Recognition, IEEE Computer Society, 2003, pp. 682-682.

  3. D. Karatzas, F. Shafait, S. Uchida, M. Iwamura, L. Gomez, S. Robles, J. Mas, D. Fernandez, J. Almazan, L.P. de las Heras , "ICDAR 2013 Robust Reading Competition", In Proc. 12th International Conference of Document Analysis and Recognition, 2013, IEEE CPS, pp. 1115-1124. [pdf] [poster] [presentation]

  4. D. Karatzas, S. Robles Mestre, J. Mas, F. Nourbakhsh, P. Pratim Roy , "ICDAR 2011 Robust Reading Competition - Challenge 1: Reading Text in Born-Digital Images (Web and Email)", In Proc. 11th International Conference of Document Analysis and Recognition, 2011, IEEE CPS, pp. 1485-1490. [pdf] [presentation]

  5. A. Shahab, F. Shafait, A. Dengel, "ICDAR 2011 Robust Reading Competition - Challenge 2: Reading Text in Scene Images",  In Proc. 11th International Conference of Document Analysis and Recognition, 2011, IEEE CPS, pp. 1491-1496. [pdf]

 

1987registered users
76countries
5867submitted methods (*)

(*) overall number of submissions including private and public ones

Important Dates
  • 15 January: Web site online
  • 15 January until 31 March: Registration of interest
  • 5 March: Datasets available
  • 2 April: Submission of results deadline
  • 23-26 August: Results presentation