tfserve_vision4.py

INPUT_SHAPE = (28, 28, 1)

def create_cnn_architecture_model1(input_shape):
    inp = keras.layers.Input(shape=input_shape)

    conv1 = keras.layers.Conv2D(filters=16, kernel_size=(3, 3), strides=(1, 1), 
                                activation='relu', padding='same')(inp)
    pool1 = keras.layers.MaxPooling2D(pool_size=(2, 2))(conv1)
    conv2 = keras.layers.Conv2D(filters=32, kernel_size=(3, 3), strides=(1, 1), 
                                   activation='relu', padding='same')(pool1)
    pool2 = keras.layers.MaxPooling2D(pool_size=(2, 2))(conv2)

    flat = keras.layers.Flatten()(pool2)

    hidden1 = keras.layers.Dense(256, activation='relu')(flat)
    drop1 = keras.layers.Dropout(rate=0.3)(hidden1)

    out = keras.layers.Dense(10, activation='softmax')(drop1)

    model = keras.Model(inputs=inp, outputs=out)
    model.compile(optimizer='adam',
                    loss='sparse_categorical_crossentropy',
                    metrics=['accuracy'])
    return model
    
model = create_cnn_architecture_model1(input_shape=INPUT_SHAPE)
model.summary()

# Output
Model: "model"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         [(None, 28, 28, 1)]       0         
_________________________________________________________________
conv2d (Conv2D)              (None, 28, 28, 16)        160       
_________________________________________________________________
...
...
dropout (Dropout)            (None, 256)               0         
_________________________________________________________________
dense_1 (Dense)              (None, 10)                2570      
=================================================================
Total params: 409,034
Trainable params: 409,034
Non-trainable params: 0
_________________________________________________________________