CASIA-10k Chinese Scene Text Dataset
CASIA-10K is a Chinese scene text dataset. This dataset contains 10000 images under various scenarios, in which 7000 images are for training and 3000 images are for testing. For each text line, 8 coordinates of a quadrilateral are annotated. In evaluation stage, line-level predictions are required.
Download CASIA-10k （5.2G）