python-kerasHow do I set the input shape when using Keras with Python?

When using Keras with Python, the input shape of a model can be set when defining the model. For example, if defining a model with two input layers:

model = Sequential()
model.add(Dense(32, input_shape=(784,)))
model.add(Dense(32))

The input_shape argument of the first layer defines the shape of the input data. In this example, the shape is (784,), which means the input is an array of 784 elements.

If a model has more than one input layer, the shape of each layer must be specified. For example:

model = Sequential()
model.add(Dense(32, input_shape=(784,)))
model.add(Dense(32, input_shape=(128,)))

The first layer has an input shape of (784,) and the second layer has an input shape of (128,).

The input_shape argument can also be a tuple of integers, such as (784, 1). This indicates that the input is an array of 784 elements, each element being an array of 1 element.

If the input data is an image, the shape of the input should include the number of channels (e.g. 3 for RGB images). For example:

model = Sequential()
model.add(Dense(32, input_shape=(128, 128, 3)))

The input shape is (128, 128, 3), which indicates that the input is an array of 128x128 pixels, with 3 channels (RGB).

For more information, see the Keras documentation.

Edit this code on GitHub