How to increase the number and size of U-net layer in Matlab?

Dear Researchers,

I would like to know how to increase the number of U-net layers and size of U-net layers. With reference to the following command (in Matlab): lgraph = unetLayers(imageSize,numClasses). How can I proceed to do that?

In Python it is easy to do it. For example,

model = tf.keras.sequential([

tf.keras.layers.Flatten(input_shape =(28,28,1)), # flattens it into a vector (we get our first flatten input)

tf.keras.layers.Dense(hidden_layer_size, activation = 'relu'), # (we get our 1st hidden layer)

tf.keras.layers.Dense(hidden_layer_size, activation = 'relu'), # (we get our 2nd hidden layer)

tf.keras.layers.Dense(output_size, activation = 'softmax'),

As can be seen, I can keep copy-pasting layers (to increase their numbers), with the option to define their size and activation function. How can I do the same in U-net, in Matlab?

Answer 1

[Viswa]

Hi,

From your question, it looks like you are looking for options to customize the U-Net network with unetLayers() function.

unetLayers creates U-Net layers for semantic segmentation. You can find more information about the function and different input arguments in the following page.

https://in.mathworks.com/help/vision/ref/unetlayers.html

You can also refer to the following page for interactively building networks with Deep Network Designer.

https://in.mathworks.com/help/deeplearning/ug/build-networks-with-deep-network-designer.html