Class MobileNetV2
java.lang.Object
ai.beestream.jitnn.nn.models.MobileNetV2
MobileNetV2 (Sandler et al., 2018), torchvision's layout: inverted residual blocks (1x1 expand,
3x3 depthwise, 1x1 linear projection) with ReLU6. The
small variant uses stride 1 in the
stem and the first stride-2 stage, for 28-32 pixel inputs.-
Method Summary
Modifier and TypeMethodDescriptionstatic SequentialmobilenetV2(int inChannels, int classes, boolean small, float dropout, long seed) MobileNetV2 over images ofinChannelschannels withclassesoutputs and dropoutdropoutbefore the classifier, initialised fromseed.
-
Method Details
-
mobilenetV2
public static Sequential mobilenetV2(int inChannels, int classes, boolean small, float dropout, long seed) MobileNetV2 over images ofinChannelschannels withclassesoutputs and dropoutdropoutbefore the classifier, initialised fromseed.
-