A Seq2seq-based Model with Global Semantic Context for Scene Text Recognition | IEEE Conference Publication | IEEE Xplore