Fig. 5From: An end-to-end text spotter with text relation networksQualitative recognition results of the “e2e-baseline” and our text spotting modelBack to article page