Key points are not available for this paper at this time.
In this paper, we present a novel SpaTial Attention Residue Network (STAR-Net) for recognising texts. The overall architecture of our STAR-Net is illustrated in fig. 1. Our STARNet emphasises the importance of representative image-based feature extraction from text regions by the spatial attention mechanism and the residue learning strategy. It is by far the deepest neural network proposed for scene text recognition. . .
Liu et al. (2016) studied this question.