Change Conv2DTranspose output shape from (None,28,28,1) to (None,32,32,1)

Question

I am trying to implement a Decoder whose output shape is (None,32,32,1). The snippet below implements a decoder whose output shape is (None,28,28,1) instead:

# Decoder
latent_dim = 2
latent_inputs = keras.Input(shape=(latent_dim,))
x = layers.Dense(7 * 7 * 64, activation="relu")(latent_inputs)
x = layers.Reshape((7, 7, 64))(x)
x = layers.Conv2DTranspose(64, 3, activation="relu", strides=2, padding="same")(x)
x = layers.Conv2DTranspose(32, 3, activation="relu", strides=2, padding="same")(x)
decoder_outputs = layers.Conv2DTranspose(1, 3, activation="sigmoid", padding="same")(x)
decoder = keras.Model(latent_inputs, decoder_outputs, name="decoder")
decoder.summary()

Summary is:

Model: "decoder"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_2 (InputLayer)        [(None, 2)]               0         
                                                                 
 dense_1 (Dense)             (None, 3136)              9408      
                                                                 
 reshape (Reshape)           (None, 7, 7, 64)          0         
                                                                 
 conv2d_transpose (Conv2DTr  (None, 14, 14, 64)        36928     
 anspose)                                                        
                                                                 
 conv2d_transpose_1 (Conv2D  (None, 28, 28, 32)        18464     
 Transpose)                                                      
                                                                 
 conv2d_transpose_2 (Conv2D  (None, 28, 28, 1)         289       
 Transpose)                                                      
                                                                 
=================================================================
Total params: 65089 (254.25 KB)
Trainable params: 65089 (254.25 KB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________

• How can I achieve that? Can you help me?

Answer 1

What about to simply replace 7 by 8?

x = layers.Dense(8 * 8 * 64, activation="relu")(latent_inputs)
x = layers.Reshape((8, 8, 64))(x)