Table 8 Comparison with transformer-based architectures.

From: Optimizing diabetic retinopathy detection with electric fish algorithm and bilinear convolutional networks

Model

Accuracy (%)

Precision (%)

Recall (%)

F1-score (%)

ViT

91.8

90.7

89.4

90.0

Swin Transformer

92.4

91.2

90.1

90.6

BCAN (proposed)

94.6

93.7

92.7

91.7

Back to article page