Untitled

def get_baseline_regression_model(input_shape, n_classes):
    input_1 = Input(shape=input_shape, name="input_1")

    conv_1 = Conv2D(96, kernel_size=11, kernel_initializer='he_normal')(input_1)
    conv_1 = BatchNormalization()(conv_1)
    conv_1 = Activation('relu')(conv_1)
    
    pool_1 = MaxPooling2D((2, 2))(conv_1)

    conv_2 = Conv2D(128, kernel_size=5, kernel_initializer='he_normal')(pool_1)
    conv_2 = BatchNormalization()(conv_2)
    conv_2 = Activation('relu')(conv_2)
    
    pool_2 = MaxPooling2D((2, 2))(conv_2)

    conv_3 = Conv2D(256, kernel_size=3, kernel_initializer='he_normal')(pool_2)
    conv_3= BatchNormalization()(conv_3)
    conv_3 = Activation('relu')(conv_3)

    conv_4 = Conv2D(256, kernel_size=3, kernel_initializer='he_normal')(conv_3)
    conv_4 = BatchNormalization()(conv_4)
    conv_4 = Activation('relu')(conv_4)

    pool_4 = MaxPooling2D((2, 2))(conv_4)

    conv_5 = Conv2D(256, kernel_size=3, kernel_initializer='he_normal')(pool_4)
    conv_5 = BatchNormalization()(conv_5)
    conv_5 = Activation('relu')(conv_5)

    pool_5 = MaxPooling2D((2, 2))(conv_5)

    conv_6 = Conv2D(256, kernel_size=3, kernel_initializer='he_normal')(pool_5)
    conv_6 = BatchNormalization()(conv_6)
    conv_6 = Activation('relu')(conv_6)

    pool_6 = MaxPooling2D((2, 2))(conv_6)

    flatten = Flatten()(pool_6)

    fc_1 = Dense(64, activation='relu')(flatten)
    fc_1 = Dropout(0.6)(fc_1)

    fc_2 = Dense(128, activation='relu')(fc_1)
    fc_2 = Dropout(0.6)(fc_2)

    out =  Dense(n_classes*2, activation='tanh')(fc_2)

    model = Model(input_1, out)
    '''
    for layer in model.layers:
        print(layer.name, layer.trainable)
    '''

    return model