Table 4 Performance comparison of different models on mobile devices.

From: A study on expression recognition based on improved mobilenetV2 network

model

Average prediction time/ms

MobileNetV1

221

MobileNetV2

300

ours

78

Back to article page