Table 11 Layer-wise statistical analysis of OCRNet.

From: OCRNet a robust deep learning framework for alphanumeric character recognition to assist the visually impaired

Model configuration

Convolution type

Accuracy

F1-score

Inference time (ms)

CNN (blocks 1–4 only)

Standard convolution

0.92

0.93

110

CNN + block 5

Standard convolution

0.93

0.935

118

CNN + blocks 5–6

Standard convolution

0.94

0.945

130

CNN + blocks 5–6 with separable convolution

Separable convolution

0.95

0.96

120