Table 11 Model performance and complexity comparison.

From: Real-time driver drowsiness detection using transformer architectures: a novel deep learning approach

Model

Accuracy (%)

Parameters

Average inference time (ms/frame)

Average FPS

ViT transformer

99.15

85,800,194

1021.83

0.98

Swin transformer

99.03

27,914,108

472.8

2.12

InceptionV3

98.5

23,867,553

136.33

7.35

InceptionResNetV2

98.5

55,851,105

122.8

8.14

MobileNet

98

3,208,001

55

18.18

Back to article page