Abstract: End-to-end scene text spotting has made significant progress due to its intrinsic synergy between text detection and recognition. Previous methods commonly regard manual annotations such as ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results