Robust ReadingCompetition
Challenges

Overview - ICDAR2017 Robust Reading Challenge on COCO-Text

This is a challenge on scene text detection and recognition, based on the largest scene text dataset currently available, based on real (as opposed to synthetic) scene imagery: the COCO-Text dataset [1]. It is structured around three tasks: Text Localization, Cropped Word Recognition and End-To-End Recognition. See detains in Tasks page.

COCO-Text is based on the MS COCO dataset, which contains images of complex everyday scenes. The images were not collected with text in mind and thus contain a broad variety of text instances. In this sense, they relate to ICDAR 2015 Robust Reading Competition (RRC) - Challenge 4, on incidental text, referring to “text that appears in the scene without the user having taken any specific prior action to cause its appearance or improve its positioning / quality in the frame.” [2].

Text in the COCO-Text dataset is annotated with (a) location in terms of a bounding box, (b) fine-grained classification into machine printed text and handwritten text, (c) classification into legible and illegible text, (d) script of the text and (e) transcriptions of legible text. The dataset contains over 173,589 labeled text regions in over 63,686 images. This signifies an order of magnitude change from the 1,500 images and 7,548 regions of the dataset of RRC 2015 - Challenge 4.

 

COCO_train2014_000000003157.jpgCOCO_train2014_000000025102.jpgCOCO_train2014_000000028392.jpg

 

References

[1]  A. Veit, T. Matera, L. Neumann, J. Matas, S. Belongie. COCO-Text: Dataset and Benchmark for Text Detection and Recognition in Natural Images. arXiv preprint arXiv:1601.07140, 2016.

[2]  D. Karatzas, L. Gomez-Bigorda, A. Nicolaou, D. Ghosh , A. Bagdanov, M. Iwamura, J. Matas, L. Neumann, VR. Chandrasekhar, A. Lu, F. Shafait, S. Uchida, E. Valveny.: ICDAR 2015 robust reading competition. 13th International Conference on Document Analysis and Recognition (ICDAR).

Important Dates
  • March, 13: COCO-Text available. (train/val/test).

  • March, 19: Cropped words dataset available. (train/val).

  • March, 23: Annotations updated (v1.4).

  • March, 30: Cropped words dataset updated (v1.4).

  • May, 23:    Submissions opening.

  • June, 30: Submission of results deadline.

  • November, 10-15: Results presentation.