base_model = MobileNetV2(input_shape=(224, 224, 3),
                                      include_top=False,
                                      weights='imagenet')

base_model.trainable = False

model = Sequential([base_model,
                    GlobalAveragePooling2D(),
                    Dense(1, activation='sigmoid')])

model.compile(loss='binary_crossentropy',
              optimizer=RMSprop(lr=0.001),
              metrics=['accuracy'])

train_datagen = ImageDataGenerator(rescale=1./255)
val_datagen = ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
        train_dir,
        target_size=(224, 224),
        batch_size=32,
        class_mode='binary')

validation_generator = val_datagen.flow_from_directory(
        validation_dir,
        target_size=(224, 224),
        batch_size=32,
        class_mode='binary')

Found 2000 images belonging to 2 classes.
Found 1000 images belonging to 2 classes.

history = model.fit(
      train_generator,
      epochs=6,
      validation_data=validation_generator,
      verbose=2)

Train for 63 steps, validate for 32 steps
Epoch 1/6
63/63 - 470s - loss: 0.3610 - accuracy: 0.8515 - val_loss: 0.1664 - val_accuracy: 0.9460
Epoch 2/6
63/63 - 139s - loss: 0.2055 - accuracy: 0.9210 - val_loss: 0.2065 - val_accuracy: 0.9150
Epoch 3/6
63/63 - 89s - loss: 0.1507 - accuracy: 0.9470 - val_loss: 0.1294 - val_accuracy: 0.9560
Epoch 4/6
63/63 - 81s - loss: 0.1342 - accuracy: 0.9510 - val_loss: 0.1733 - val_accuracy: 0.9350
Epoch 5/6
63/63 - 80s - loss: 0.1205 - accuracy: 0.9555 - val_loss: 0.1684 - val_accuracy: 0.9390
Epoch 6/6
63/63 - 82s - loss: 0.1079 - accuracy: 0.9620 - val_loss: 0.1418 - val_accuracy: 0.9450