The ARText dataset contains 33,000 artistic images with rich shape deformation and font diversity, and has been randomly divided into training set (30,000 images) and testing set (3,000 images) for model development.
BibTex:
Before browse our site, please accept our cookies policy
