Text Recognition in Natural Scenes

From TC11
Jump to: navigation, search

Datasets -> Datasets List -> Current Page

Created: 2010-10-03
Last updated: 2011-10-03

Proposed By

Robert Nagy
University of Erlangen-Nuremberg
Chair for Computer Science 6 (Data Management)
Matrensstr. 3
D-91058 Erlangen
Email: robert[dot]nagy [at] cs[dot]fau[dot]de


Text recognized in natural scenes can provide information and help for visually impaired, mobile applications, navigation systems, object classification and image annotation. Unfortunately, current OCR systems and methods recognize text in natural scenes very poorly.

Training and test image lists are provided along with the dataset. For measuring the average similarity of recognized and ground truth text, a normalized similarity measure is defined in the corresponding technical report. The aim is not to measure the pixel coverage of detected text, but rather the recognized text itself, because only this higher level can be used in the application domains mentioned above.

Related Dataset and Ground Truth Data

Related Software



  1. R. Nagy, A. Dicker and K. Meyer‐Wegener, "NEOCR: A Configurable Dataset for Natural Image Text Recognition". In CBDAR Workshop 2011 at ICDAR 2011. pp. 53‐58, September 2011. (PDF), (Presentation)
  2. R. Nagy, A. Dicker, and K. Meyer‐Wegener, "Definition and Evaluation of the NEOCR Dataset for Natural‐Image Text Recognition". University of Erlangen, Dept. of Computer Science, Technical Reports, CS‐2011‐07, September 2011. (PDF)

This page is editable only by TC11 Officers .