How can I freeze last layer of my own model?

Question

model = ResNet50(weights='imagenet', include_top=False, input_shape=(224, 224, 3))

The code above allows me to use weights of imagenet but i want to use my own weights insted of imagenet, what changes should I make in my code that will allow me to train just last layer on my own dataset? Here is the code for my model:

def mini_XCEPTION(input_shape, num_classes, l2_regularization=0.01):
    regularization = l2(l2_regularization)

    # base
    img_input = Input(input_shape)
    x = Conv2D(8, (3, 3), strides=(1, 1), kernel_regularizer=regularization,
               use_bias=False)(img_input)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(8, (3, 3), strides=(1, 1), kernel_regularizer=regularization,
               use_bias=False)(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)

    # module 1
    residual = Conv2D(16, (1, 1), strides=(2, 2),
                      padding='same', use_bias=False)(x)
    residual = BatchNormalization()(residual)

    x = SeparableConv2D(16, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = SeparableConv2D(16, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)

    x = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(x)
    x = layers.add([x, residual])

    # module 2
    residual = Conv2D(32, (1, 1), strides=(2, 2),
                      padding='same', use_bias=False)(x)
    residual = BatchNormalization()(residual)

    x = SeparableConv2D(32, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = SeparableConv2D(32, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)

    x = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(x)
    x = layers.add([x, residual])

    # module 3
    residual = Conv2D(64, (1, 1), strides=(2, 2),
                      padding='same', use_bias=False)(x)
    residual = BatchNormalization()(residual)

    x = SeparableConv2D(64, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = SeparableConv2D(64, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)

    x = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(x)
    x = layers.add([x, residual])

    # module 4
    residual = Conv2D(128, (1, 1), strides=(2, 2),
                      padding='same', use_bias=False)(x)
    residual = BatchNormalization()(residual)

    x = SeparableConv2D(128, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = SeparableConv2D(128, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
    x = BatchNormalization()(x)

    x = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(x)
    x = layers.add([x, residual])

    x = Conv2D(num_classes, (3, 3),
               # kernel_regularizer=regularization,
               padding='same')(x)
    x = GlobalAveragePooling2D()(x)
    output = Activation('softmax', name='predictions')(x)

    model = Model(img_input, output)
    return model

You need to tag or at least mention what third-party module(s) you're using when asking questions here — I made some guesses... — martineau, Mar 09 '19 at 13:11

score 0 · Answer 1 · answered Mar 09 '19 at 13:27

First, you load the saved weights for ResNet50. After that, you use the same architecture and in the last layer, you do:

last_layer=GlobalAveragePooling2D()(x)
last_layer.trainable=True

For all previous layers, you make trainable=False, like:

x = Conv2D(8, (3, 3), strides=(1, 1), kernel_regularizer=regularization,
               use_bias=False)(img_input)
x.trainable=False

residual = Conv2D(16, (1, 1), strides=(2, 2),
                      padding='same', use_bias=False)(x)
residual.trainable=False

residual = BatchNormalization()(residual)
residual.trainable=False

x = SeparableConv2D(16, (3, 3), padding='same',
                        kernel_regularizer=regularization,
                        use_bias=False)(x)
x.trainable=False

Resnet50 is built in keras model, but the code(mini_XCEPTION) i provided has no option(arguments) to use weights, kindly tell me what changes should i make in above code that will allow me to use previous weights. After which i could perform the steps you mentioned. — Zunaira Akmal, Mar 10 '19 at 09:55
Please check this link Zunaira: https://stackoverflow.com/questions/35074549/how-to-load-a-model-from-an-hdf5-file-in-keras — razimbres, Mar 10 '19 at 10:03

How can I freeze last layer of my own model?

1 Answers1