CODEWITHSUNDEEP
tensorflowkeras
HuckleberryFinn
HuckleberryFinn

Reputation: 1529

Transposed convolution in tf.keras.layers

How can I convert this function to an equivalent tf.keras (Tensorflow 2.0) implementation?

def deconv2d(inputs,out_channels, name=None):
batch, in_height, in_width, in_channels = [int(d) for d in inputs.get_shape()]
filter = tf.get_variable(name + "filter", [4, 4, out_channels, in_channels], initializer= tf.variance_scaling_initializer())
return tf.nn.conv2d_transpose(inputs, filter, [batch, in_height * 2, in_width * 2, out_channels], strides=[1,2,2,1],
                             padding='SAME', name= name)

Upvotes: 1

Views: 328

Answers (1)

Sharky
Sharky

Reputation: 4533

TensoFlow 2.0 has tf.keras.layers.Conv2DTranspose. Keras is the default high level API for TF 2.0, but it still have tf.nn.conv2d_transpose for low level applications

https://www.tensorflow.org/versions/r2.0/api_docs/python/tf/keras/layers/Conv2DTranspose

Upvotes: 1

Related Questions