testModel.py

# coding: utf-8

# In[18]:


import numpy as np
from keras import layers
from keras.layers import Input, Dense, Activation, ZeroPadding2D, BatchNormalization, Flatten, Conv2D
from keras.layers import AveragePooling2D, MaxPooling2D, Dropout, GlobalMaxPooling2D, GlobalAveragePooling2D
from keras.models import Model
from keras.preprocessing import image
from keras.utils import layer_utils
from keras.utils.data_utils import get_file
from keras.applications.imagenet_utils import preprocess_input
import pydot
import keras
from IPython.display import SVG
from keras.utils.vis_utils import model_to_dot
from keras.utils import plot_model

from keras.applications.mobilenet import MobileNet

import matplotlib.pyplot as plt
get_ipython().run_line_magic('matplotlib', 'inline')

import keras.backend as K
K.set_image_data_format('channels_last')
K.set_learning_phase(1)

from keras.models import model_from_json
from keras.utils.generic_utils import CustomObjectScope


# In[35]:


def myPredict(img_path, model):
    img = image.load_img(img_path, target_size=(128, 128))
    x = image.img_to_array(img)
    x = np.expand_dims(x, axis=0)
    x = preprocess_input(x)
    return model.predict(x)[0]


# In[28]:


import tensorflow as tf
from keras.backend.tensorflow_backend import set_session
config = tf.ConfigProto()
config.gpu_options.allow_growth=True
set_session(tf.Session(config=config))


# In[30]:


from keras.models import load_model
model = load_model('flower_model.h5', custom_objects={
                   'relu6': keras.applications.mobilenet.relu6,
                   'DepthwiseConv2D': keras.applications.mobilenet.DepthwiseConv2D})


# In[36]:


preds = myPredict("test-tulips.jpg", model)
result = np.array(preds).argmax()
print(result) # ans: 1

transfer.py

# coding: utf-8

# In[1]:


import numpy as np
from keras import layers
from keras.layers import Input, Dense, Activation, ZeroPadding2D, BatchNormalization, Flatten, Conv2D
from keras.layers import AveragePooling2D, MaxPooling2D, Dropout, GlobalMaxPooling2D, GlobalAveragePooling2D
from keras.models import Model
from keras.preprocessing import image
from keras.utils import layer_utils
from keras.utils.data_utils import get_file
from keras.applications.imagenet_utils import preprocess_input
import pydot
from IPython.display import SVG
from keras.utils.vis_utils import model_to_dot
from keras.utils import plot_model

import tensorflow as tf
from keras.backend.tensorflow_backend import set_session
config = tf.ConfigProto()
config.gpu_options.allow_growth=True
set_session(tf.Session(config=config))


# In[2]:


train_dir = "flower_photos_small/train"
test_dir = "flower_photos_small/test"

IM_WIDTH = 128
IM_HEIGHT = 128
batch_size = 36


# In[3]:


from keras.preprocessing.image import ImageDataGenerator

train_datagen =  ImageDataGenerator(
    preprocessing_function=preprocess_input,
    rotation_range=30,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True
)

test_datagen = ImageDataGenerator(
    preprocessing_function=preprocess_input,
    rotation_range=30,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True
)


train_generator = train_datagen.flow_from_directory(
  train_dir,
  target_size=(IM_WIDTH, IM_HEIGHT),
  batch_size=batch_size,
)

validation_generator = test_datagen.flow_from_directory(
  test_dir,
  target_size=(IM_WIDTH, IM_HEIGHT),
  batch_size=batch_size,
)


# In[4]:


from keras.applications.mobilenet import MobileNet

base_model = MobileNet(weights='imagenet', include_top=False, input_shape=(128, 128, 3))


# In[5]:


base_model.summary()


# In[6]:


def add_new_layer(base_model, classes):
    x = base_model.output
    x = GlobalAveragePooling2D()(x)
    x = Dense(36, activation='relu')(x)
    predictions = Dense(classes, activation='softmax')(x) 
    model = Model(input=base_model.input, output=predictions)
    return model


# In[7]:


model = add_new_layer(base_model, 5)

base_model.trainable = False
model.compile(optimizer='adam',
            loss='categorical_crossentropy',
            metrics=['accuracy'])


# In[8]:


history = model.fit_generator(
      train_generator,
      steps_per_epoch=100,
      epochs=20,
      validation_data=validation_generator,
      validation_steps=30)


# In[9]:


model.save("flower_model.h5")


# In[10]:


model.save_weights("flower_model_weights.h5")


# In[11]:


jsonstr = model.to_json()
f = open("flower_model_json.json", "w")
f.write(jsonstr)
f.close()


# In[26]:


def myPredict(img_path, model):
    img = image.load_img(img_path, target_size=(128, 128))
    x = image.img_to_array(img)
    x = np.expand_dims(x, axis=0)
    x = preprocess_input(x)
    return model.predict(x)[0]


# In[27]:


preds = myPredict("test-tulips.jpg", model)
result = np.array(preds).argmax()
print(result) # ans: 4