Giters

cs-chan / Total-Text-Dataset

Total Text Dataset. It consists of 1555 images with more than 3 different text orientations: Horizontal, Multi-Oriented, and Curved, one of a kind.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The result of end-to-end recognition.

MhLiao opened this issue · comments

commented

@cs-chan It seems that there is a mistake in the result of end-to-end recognition.
In the Neurocomputing2019 paper, the authors say that "Total-Text [33] contains 744 horizontal text images for testing", which means that they did not evaluate the curved samples. Moreover, it is only a text recognition method which does not involve detection, instead of an end-to-end recognition method which involves both detection and recognition.

commented

Thank you for your information. We have checked and confirmed it. In this case, we have removed the paper in the table to avoid confusion.