Skip to main content

Table 1 The CNN architecture

From: Large scale tissue histopathology image classification, segmentation, and visualization via deep convolutional activation features

Layer Dimension Kernel size Stride Details
input 224×224×3 - - RGB channels
conv1 55×55×96 11 4 -
pool1 27×27×96 3 2 Max pooling
conv2 27×27×256 5 1 -
pool2 13×13×256 3 2 Max pooling
conv3 13×13×384 3 1 -
conv4 13×13×384 3 1 -
conv5 13×13×256 3 1 -
pool3 6×6×256 3 2 Max pooling
fc1 4096 - - -
fc2 4096 - - -