From:  Artificial intelligence fusion for predicting survival of rectal cancer patients using immunohistochemical expression of Ras homolog family member B in biopsy
 

Properties of pretrained networks

Pretrained CNN modelLayersModel size (MB)ParametersInput size (pixels)
ResNet-18184411.7 × 106224 × 224
SqueezeNet185.21.24 × 106227 × 227
DenseNet-2012017720 × 106224 × 224
AlexNet822761 × 106227 × 227
Xception718522.9 × 106299 × 299
NASNet-Large1244#33288.9 × 106331 × 331
Return to article view

indicates the network does not consist of a linear sequence of modules; NASNet: neural search architecture network; ResNet: residual network; DenseNet: dense convolutional network