HRNet Encoder and Dual-Branch Decoder Framework-Based Scene Text Recognition Model

Figure 8

Comparison on the feature maps generated by the Independent Trans Conv2D layers and bilinear interpolation.