FDTA: Fully Convolutional Scene Text Detection With Text Attention | IEEE Journals & Magazine | IEEE Xplore