Skip to main content

Table 3 Backbone performance comparisons in the test set

From: SwinOCSR: end-to-end optical chemical structure recognition using a Swin Transformer

Backbone

Accuracy

Tanimoto

BLEU

ROUGE

Swin Transformer(our)

0.9736

0.9965

0.9946

0.9964

ResNet-50

0.8917

0.9879

0.9862

0.9887

EfficientNet-B3

0.8670

0.9846

0.9837

0.9866