COO: Comic Onomatopoeia Dataset for Recognizing Arbitrary or Truncated Texts.

Jeonghun Baek, Yusuke Matsui, Kiyoharu Aizawa

Browse the full ECCV paper archive.