Synthetic Data for Text Localisation in Natural Images

About

In this paper we introduce a new method for text detection in natural images. The method comprises two contributions: First, a fast and scalable engine to generate synthetic images of text in clutter. This engine overlays synthetic text to existing background images in a natural way, accounting for the local 3D scene geometry. Second, we use the synthetic images to train a Fully-Convolutional Regression Network (FCRN) which efficiently performs text detection and bounding-box regression at all locations and multiple scales in an image. We discuss the relation of FCRN to the recently-introduced YOLO detector, as well as other end-to-end object detection systems based on deep learning. The resulting detection network significantly out performs current methods for text detection in natural images, achieving an F-measure of 84.2% on the standard ICDAR 2013 benchmark. Furthermore, it can process 15 images per second on a GPU.

Ankush Gupta, Andrea Vedaldi, Andrew Zisserman · 2016

Related benchmarks

Task	Dataset	Result
Text Detection	ICDAR 2013 (test)	F1 Score83	88
Text Localization	ICDAR 2013 (test)	Recall76	28
End-to-End Text Spotting	ICDAR 2013 (test)	Score S—	25
End-to-End Text Spotting	ICDAR 2011 (test)	F-measure84.3	12
End-to-end Recognition	ICDAR 2013	Strong F-Measure85	8
End-to-End Text Spotting	Street View Text (SVT)	Max F-measure55.7	7
End-to-End Text Spotting	Street View Text SVT-50 Constrained lexicon	Maximum F1-Score68	7
Word Spotting	ICDAR 2013 (test)	Generic Metric (G)84.7	6