def InceptionV3(nbclasses):
    # use: model = InceptionV3()
    base_model = keras.applications.inception_v3.InceptionV3(include_top=False, weights='imagenet')
    x = base_model.output
    x = GlobalAveragePooling2D(name='avg_pool')(x)
    x = Dropout(0.4)(x)
    predictions = Dense(nbclasses, activation='softmax')(x)
    model = Model(inputs=base_model.input, outputs=predictions)

    for layer in base_model.layers:
        layer.trainable = False

    model.compile(optimizer='rmsprop',
                  loss='categorical_crossentropy',
                  metrics=['accuracy'])

    return model


class InceptionV3:
    # use:
    # inc = InceptionV3()
    # model = inc.build()
    def __init__(self, nbclasses, weights='imagenet'):
        base_model = keras.applications.inception_v3.InceptionV3(include_top=False, weights=weights)
        x = base_model.output
        x = GlobalAveragePooling2D(name='avg_pool')(x)
        x = Dropout(0.4)(x)
        predictions = Dense(nbclasses, activation='softmax')(x)
        self.model = Model(inputs=base_model.input, outputs=predictions)

        for layer in base_model.layers:
            layer.trainable = False

    def build(self, metrics=None):
        if not metrics:
            metrics = ['accuracy']
        self.model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=metrics)
        return self.model


class InceptionV3:
    # use: model = InceptionV3.build()
    @classmethod
    def build(cls, nbclasses, weights='imagenet'):
        base_model = keras.applications.inception_v3.InceptionV3(include_top=False, weights=weights)
        x = base_model.output
        x = GlobalAveragePooling2D(name='avg_pool')(x)
        x = Dropout(0.4)(x)
        predictions = Dense(nbclasses, activation='softmax')(x)
        model = Model(inputs=base_model.input, outputs=predictions)

        for layer in base_model.layers:
            layer.trainable = False

        model.compile(optimizer='rmsprop',
                      loss='categorical_crossentropy',
                      metrics=['accuracy'])

        return model