import cv2
import keras
import numpy as np
from keras import backend as K, Sequential
from keras.layers import Conv2D, Flatten, Dense, Dropout, Activation, GlobalAveragePooling2D
from keras.optimizers import Adam
from keras.preprocessing.image import ImageDataGenerator
from matplotlib import pyplot as plt
# %matplotlib inline

TRAIN_DIR = './images/train'
TEST_DIR = './images/test'
VALID_DIR = './images/valid'
IMAGE_CLASS = ['Anggur', 'Apel', 'ApelH', 'Jeruk', 'Kiwi', 'Lemon', 'Mangga', 'Melon', 'Naga', 'Pisang', 'Salak',
               'Sirsak']

TRAIN_BATCH = ImageDataGenerator().flow_from_directory(TRAIN_DIR, target_size=(224, 224), classes=IMAGE_CLASS,
                                                       batch_size=10)
TEST_BATCH = ImageDataGenerator().flow_from_directory(TEST_DIR, target_size=(224, 224), classes=IMAGE_CLASS,
                                                      batch_size=10)
VALID_BATCH = ImageDataGenerator().flow_from_directory(VALID_DIR, target_size=(224, 224), classes=IMAGE_CLASS,
                                                       batch_size=10)

vgg16_model = keras.applications.vgg16.VGG16()

model = Sequential()
for layer in vgg16_model.layers:
    model.add(layer)

model.layers.pop()

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

model.add(Dense(12, activation='softmax'))
# model.summary()
model.compile(Adam(lr=.0001), loss='categorical_crossentropy', metrics=['accuracy'])
model.fit_generator(TRAIN_BATCH, steps_per_epoch=15, validation_data=VALID_BATCH, validation_steps=15, epochs=100, verbose=1)