MEAN: Multi - Element Attention Network for Scene Text Recognition | IEEE Conference Publication | IEEE Xplore