Scene Text Recognition

Although the vertical Chinese text recognition dataset presented by Yu is public, it is reproduced from the PosterErase dataset, collected from the e-commerce platform for the poster text erasing task, and does not contain the challenges from real application scenarios. Therefore, we establish a benchmark dataset (Vertical and Horizontal Text Recognition Dataset, WHU-VHTR) to promote in-depth research on STR. WHU-VHTR contained 23674 images annotated with line-level transcriptions, collecting from Google Street View and real urban scene images in China.


Scene text detection and recognition have attracted much attention in recent years because of their potential applications. Detecting and recognizing texts in images may suffer from scene complexity and text variations. Some of these problematic cases are included in popular benchmark datasets, but only to a limited extent. In this work, we investigate the problem of scene text detection and recognition in a domain with extreme challenges.