Keras กวดวิชา

from keras.models import Sequential
from keras.layers import Dense, Activation

model = Sequential()
model.add(Dense(64, activation='relu', input_dim=50)) #input shape of 50
model.add(Dense(28, activation='relu')) #input shape of 50
model.add(Dense(10, activation='softmax'))

pip install --upgrade tensorflow

pip install --upgrade tensorflow-gpu

user@user:~$ python
Python 3.6.4 (default, Mar 20 2018, 11:10:20) 
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
>>>

pip install keras

gedit ~/.keras/keras.json

{
    "floatx": "float32",
    "epsilon": 1e-07,
    "backend": "tensorflow",
    "image_data_format": "channels_last"
}

user@user:~$ python3
Python 3.6.4 (default, Mar 20 2018, 11:10:20) 
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import keras
Using TensorFlow backend.
>>> 

pip3 install –U pip virtualenv

pip3 install virtualenv

virtualenv –-system-site-packages –p python3 ./venv

.\venv\Scripts\activate

from keras.models import Sequential
from keras.layers import Dense, Activation,Conv2D,MaxPooling2D,Flatten,Dropout

model = Sequential()

input_shape=(320,320,3) #this is the input shape of an image 320x320x3
model.add(Conv2D(48, (3, 3), activation='relu', input_shape= input_shape))

model.add(Conv2D(48, (3, 3), activation='relu'))

model.add(MaxPooling2D(pool_size=(2, 2)))

model.add(Dense(256, activation='relu'))

model.add(Dropout(0.5))

model.compile(loss='mean_squared_error', optimizer='adam')

model.fit(X_train, X_train, batch_size=32, epochs=10, validation_data=(x_val, y_val))

score = model.evaluate(x_test, y_test, batch_size=32)

import keras
from keras.models import Sequential
from keras.layers import Dense, Activation
import numpy as np
import matplotlib.pyplot as plt 
 
x = data = np.linspace(1,2,200)
y = x*4 + np.random.randn(*x.shape) * 0.3


model = Sequential()
model.add(Dense(1, input_dim=1, activation='linear'))

model.compile(optimizer='sgd', loss='mse', metrics=['mse'])

weights = model.layers[0].get_weights()
w_init = weights[0][0][0]
b_init = weights[1][0]
print('Linear regression model is initialized with weights w: %.2f, b: %.2f' % (w_init, b_init)) 


model.fit(x,y, batch_size=1, epochs=30, shuffle=False)

weights = model.layers[0].get_weights()
w_final = weights[0][0][0]
b_final = weights[1][0]
print('Linear regression model is trained to have weight w: %.2f, b: %.2f' % (w_final, b_final))

predict = model.predict(data)

plt.plot(data, predict, 'b', data , y, 'k.')
plt.show()

Linear regression model is initialized with weights w: 0.37, b: 0.00

Linear regression model is trained to have weight w: 3.70, b: 0.61

!wget https://s3.us-east-2.amazonaws.com/naturalimages02/images.tar.gz		
!tar -xzf images.tar.gz

from keras.preprocessing.image import ImageDataGenerator
import numpy as np
import matplotlib.pyplot as plt

train_path = 'images/train/'
test_path = 'images/test/'
batch_size = 16
image_size = 224
num_class = 8


train_datagen = ImageDataGenerator(validation_split=0.3,
                                   shear_range=0.2,
                                   zoom_range=0.2,
                                   horizontal_flip=True)

train_generator = train_datagen.flow_from_directory(
                        directory=train_path,
                        target_size=(image_size,image_size),
                        batch_size=batch_size,
                        class_mode='categorical',
                        color_mode='rgb',
                        shuffle=True)

x_batch, y_batch = train_generator.next()

fig=plt.figure()
columns = 4
rows = 4
for i in range(1, columns*rows):
    num = np.random.randint(batch_size)
    image = x_batch[num].astype(np.int)
    fig.add_subplot(rows, columns, i)
    plt.imshow(image)
plt.show()

import keras
from keras.models import Model, load_model
from keras.layers import Activation, Dropout, Flatten, Dense
from keras.preprocessing.image import ImageDataGenerator
from keras.applications.vgg16 import VGG16


#Load the VGG model
base_model = VGG16(weights='imagenet', include_top=False, input_shape=(image_size, image_size, 3))

print(base_model.summary())

    # Freeze the layers 
for layer in base_model.layers:
    layer.trainable = False
 
# # Create the model
model = keras.models.Sequential()

# # Add the vgg convolutional base model
model.add(base_model)
 
# # Add new layers
model.add(Flatten())
model.add(Dense(1024, activation='relu'))
model.add(Dense(1024, activation='relu'))
model.add(Dense(num_class, activation='softmax'))
 
# # Show a summary of the model. Check the number of trainable parameters    
print(model.summary())

Layer (type)                 Output Shape              Param #   
=================================================================
vgg16 (Model)                (None, 7, 7, 512)         14714688  
_________________________________________________________________
flatten_1 (Flatten)          (None, 25088)             0         
_________________________________________________________________
dense_1 (Dense)              (None, 1024)              25691136  
_________________________________________________________________
dense_2 (Dense)              (None, 1024)              1049600   
_________________________________________________________________
dense_3 (Dense)              (None, 8)                 8200      
=================================================================
Total params: 41,463,624
Trainable params: 26,748,936
Non-trainable params: 14,714,688

# # Compile the model
from keras.optimizers import SGD

model.compile(loss='categorical_crossentropy',
          optimizer=SGD(lr=1e-3),
          metrics=['accuracy'])

# # Start the training process
# model.fit(x_train, y_train, validation_split=0.30, batch_size=32, epochs=50, verbose=2)

# # #save the model
# model.save('catdog.h5')

history = model.fit_generator(
        train_generator,
        steps_per_epoch=train_generator.n/batch_size,
        epochs=10)
        
model.save('fine_tune.h5')

# summarize history for accuracy
import matplotlib.pyplot as plt

plt.plot(history.history['loss'])
plt.title('loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['loss'], loc='upper left')
plt.show()

Epoch 1/10
432/431 [==============================] - 53s 123ms/step - loss: 0.5524 - acc: 0.9474 
Epoch 2/10
432/431 [==============================] - 52s 119ms/step - loss: 0.1571 - acc: 0.9831
Epoch 3/10
432/431 [==============================] - 51s 119ms/step - loss: 0.1087 - acc: 0.9871
Epoch 4/10
432/431 [==============================] - 51s 119ms/step - loss: 0.0624 - acc: 0.9926
Epoch 5/10
432/431 [==============================] - 51s 119ms/step - loss: 0.0591 - acc: 0.9938
Epoch 6/10
432/431 [==============================] - 51s 119ms/step - loss: 0.0498 - acc: 0.9936
Epoch 7/10
432/431 [==============================] - 51s 119ms/step - loss: 0.0403 - acc: 0.9958
Epoch 8/10
432/431 [==============================] - 51s 119ms/step - loss: 0.0248 - acc: 0.9959
Epoch 9/10
432/431 [==============================] - 51s 119ms/step - loss: 0.0466 - acc: 0.9942
Epoch 10/10
432/431 [==============================] - 52s 120ms/step - loss: 0.0338 - acc: 0.9947

model = load_model('fine_tune.h5')

test_datagen = ImageDataGenerator()
train_generator = train_datagen.flow_from_directory(
                        directory=train_path,
                        target_size=(image_size,image_size),
                        batch_size=batch_size,
                        class_mode='categorical',
                        color_mode='rgb',
                        shuffle=True)

test_generator = test_datagen.flow_from_directory(
                        directory=test_path, 
                        target_size=(image_size, image_size),
                        color_mode='rgb',
                        shuffle=False,
                        class_mode='categorical',
                        batch_size=1)

filenames = test_generator.filenames
nb_samples = len(filenames)

fig=plt.figure()
columns = 4
rows = 4
for i in range(1, columns*rows -1):
    x_batch, y_batch = test_generator.next()

    name = model.predict(x_batch)
    name = np.argmax(name, axis=-1)
    true_name = y_batch
    true_name = np.argmax(true_name, axis=-1)

    label_map = (test_generator.class_indices)
    label_map = dict((v,k) for k,v in label_map.items()) #flip k,v
    predictions = [label_map[k] for k in name]
    true_value = [label_map[k] for k in true_name]

    image = x_batch[0].astype(np.int)
    fig.add_subplot(rows, columns, i)
    plt.title(str(predictions[0]) + ':' + str(true_value[0]))
    plt.imshow(image)
plt.show()

from keras.applications.vgg16 import VGG16
from keras_vggface.vggface import VGGFace

face_model = VGGFace(model='vgg16', 
                weights='vggface',
                input_shape=(224,224,3)) 
face_model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         (None, 224, 224, 3)       0         
_________________________________________________________________
conv1_1 (Conv2D)             (None, 224, 224, 64)      1792      
_________________________________________________________________
conv1_2 (Conv2D)             (None, 224, 224, 64)      36928     
_________________________________________________________________
pool1 (MaxPooling2D)         (None, 112, 112, 64)      0         
_________________________________________________________________
conv2_1 (Conv2D)             (None, 112, 112, 128)     73856     
_________________________________________________________________
conv2_2 (Conv2D)             (None, 112, 112, 128)     147584    
_________________________________________________________________
pool2 (MaxPooling2D)         (None, 56, 56, 128)       0         
_________________________________________________________________
conv3_1 (Conv2D)             (None, 56, 56, 256)       295168    
_________________________________________________________________
conv3_2 (Conv2D)             (None, 56, 56, 256)       590080    
_________________________________________________________________
conv3_3 (Conv2D)             (None, 56, 56, 256)       590080    
_________________________________________________________________
pool3 (MaxPooling2D)         (None, 28, 28, 256)       0         
_________________________________________________________________
conv4_1 (Conv2D)             (None, 28, 28, 512)       1180160   
_________________________________________________________________
conv4_2 (Conv2D)             (None, 28, 28, 512)       2359808   
_________________________________________________________________
conv4_3 (Conv2D)             (None, 28, 28, 512)       2359808   
_________________________________________________________________
pool4 (MaxPooling2D)         (None, 14, 14, 512)       0         
_________________________________________________________________
conv5_1 (Conv2D)             (None, 14, 14, 512)       2359808   
_________________________________________________________________
conv5_2 (Conv2D)             (None, 14, 14, 512)       2359808   
_________________________________________________________________
conv5_3 (Conv2D)             (None, 14, 14, 512)       2359808   
_________________________________________________________________
pool5 (MaxPooling2D)         (None, 7, 7, 512)         0         
_________________________________________________________________
flatten (Flatten)            (None, 25088)             0         
_________________________________________________________________
fc6 (Dense)                  (None, 4096)              102764544 
_________________________________________________________________
fc6/relu (Activation)        (None, 4096)              0         
_________________________________________________________________
fc7 (Dense)                  (None, 4096)              16781312  
_________________________________________________________________
fc7/relu (Activation)        (None, 4096)              0         
_________________________________________________________________
fc8 (Dense)                  (None, 2622)              10742334  
_________________________________________________________________
fc8/softmax (Activation)     (None, 2622)              0         
=================================================================
Total params: 145,002,878
Trainable params: 145,002,878
Non-trainable params: 0
_________________________________________________________________
Traceback (most recent call last):

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

from keras.models import Model, Sequential
from keras.layers import Input, Convolution2D, ZeroPadding2D, MaxPooling2D, Flatten, Dense, Dropout, Activation

person_count = 5

last_layer = face_model.get_layer('pool5').output

x = Flatten(name='flatten')(last_layer)
x = Dense(1024, activation='relu', name='fc6')(x)
x = Dense(1024, activation='relu', name='fc7')(x)
out = Dense(person_count, activation='softmax', name='fc8')(x)

custom_face = Model(face_model.input, out)

Layer (type)                 Output Shape              Param #
=================================================================
input_1 (InputLayer)         (None, 224, 224, 3)       0
_________________________________________________________________
conv1_1 (Conv2D)             (None, 224, 224, 64)      1792
_________________________________________________________________
conv1_2 (Conv2D)             (None, 224, 224, 64)      36928
_________________________________________________________________
pool1 (MaxPooling2D)         (None, 112, 112, 64)      0
_________________________________________________________________
conv2_1 (Conv2D)             (None, 112, 112, 128)     73856
_________________________________________________________________
conv2_2 (Conv2D)             (None, 112, 112, 128)     147584
_________________________________________________________________
pool2 (MaxPooling2D)         (None, 56, 56, 128)       0
_________________________________________________________________
conv3_1 (Conv2D)             (None, 56, 56, 256)       295168
_________________________________________________________________
conv3_2 (Conv2D)             (None, 56, 56, 256)       590080
_________________________________________________________________
conv3_3 (Conv2D)             (None, 56, 56, 256)       590080
_________________________________________________________________
pool3 (MaxPooling2D)         (None, 28, 28, 256)       0
_________________________________________________________________
conv4_1 (Conv2D)             (None, 28, 28, 512)       1180160
_________________________________________________________________
conv4_2 (Conv2D)             (None, 28, 28, 512)       2359808
_________________________________________________________________
conv4_3 (Conv2D)             (None, 28, 28, 512)       2359808
_________________________________________________________________
pool4 (MaxPooling2D)         (None, 14, 14, 512)       0
_________________________________________________________________
conv5_1 (Conv2D)             (None, 14, 14, 512)       2359808
_________________________________________________________________
conv5_2 (Conv2D)             (None, 14, 14, 512)       2359808
_________________________________________________________________
conv5_3 (Conv2D)             (None, 14, 14, 512)       2359808
_________________________________________________________________
pool5 (MaxPooling2D)         (None, 7, 7, 512)         0
_________________________________________________________________
flatten (Flatten)            (None, 25088)             0
_________________________________________________________________
fc6 (Dense)                  (None, 1024)              25691136
_________________________________________________________________
fc7 (Dense)                  (None, 1024)              1049600
_________________________________________________________________
fc8 (Dense)                  (None, 5)                 5125
=================================================================
Total params: 41,460,549
Trainable params: 26,745,861
Non-trainable params: 14,714,688

from keras.preprocessing.image import ImageDataGenerator
batch_size = 5
train_path = 'data/'
eval_path = 'eval/'

train_datagen = ImageDataGenerator(rescale=1./255,
                                   shear_range=0.2,
                                   zoom_range=0.2,
                                   horizontal_flip=True)

valid_datagen = ImageDataGenerator(rescale=1./255,
                                   shear_range=0.2,
                                   zoom_range=0.2,
                                   horizontal_flip=True)

train_generator = train_datagen.flow_from_directory(
                        train_path,
                        target_size=(image_size,image_size),
                        batch_size=batch_size,
                        class_mode='sparse',
                        color_mode='rgb')

valid_generator = valid_datagen.flow_from_directory(
    directory=eval_path,
    target_size=(224, 224),
    color_mode='rgb',
    batch_size=batch_size,
    class_mode='sparse',
    shuffle=True,
)

from keras.optimizers import SGD

custom_face.compile(loss='sparse_categorical_crossentropy',
                         optimizer=SGD(lr=1e-4, momentum=0.9),
                         metrics=['accuracy'])

history = custom_face.fit_generator(
        train_generator,
        validation_data=valid_generator,
        steps_per_epoch=49/batch_size,
        validation_steps=valid_generator.n,
        epochs=50)

custom_face.evaluate_generator(generator=valid_generator)
        
custom_face.save('vgg_face.h5')
Epoch 25/50
10/9 [==============================] - 60s 6s/step - loss: 1.4882 - acc: 0.8998 - val_loss: 1.5659 - val_acc: 0.5851
Epoch 26/50
10/9 [==============================] - 59s 6s/step - loss: 1.4882 - acc: 0.8998 - val_loss: 1.5638 - val_acc: 0.5809
Epoch 27/50
10/9 [==============================] - 60s 6s/step - loss: 1.4779 - acc: 0.8597 - val_loss: 1.5613 - val_acc: 0.5477
Epoch 28/50
10/9 [==============================] - 60s 6s/step - loss: 1.4755 - acc: 0.9199 - val_loss: 1.5576 - val_acc: 0.5809
Epoch 29/50
10/9 [==============================] - 60s 6s/step - loss: 1.4794 - acc: 0.9153 - val_loss: 1.5531 - val_acc: 0.5892
Epoch 30/50
10/9 [==============================] - 60s 6s/step - loss: 1.4714 - acc: 0.8953 - val_loss: 1.5510 - val_acc: 0.6017
Epoch 31/50
10/9 [==============================] - 60s 6s/step - loss: 1.4552 - acc: 0.9199 - val_loss: 1.5509 - val_acc: 0.5809
Epoch 32/50
10/9 [==============================] - 60s 6s/step - loss: 1.4504 - acc: 0.9199 - val_loss: 1.5492 - val_acc: 0.5975
Epoch 33/50
10/9 [==============================] - 60s 6s/step - loss: 1.4497 - acc: 0.8998 - val_loss: 1.5490 - val_acc: 0.5851
Epoch 34/50
10/9 [==============================] - 60s 6s/step - loss: 1.4453 - acc: 0.9399 - val_loss: 1.5529 - val_acc: 0.5643
Epoch 35/50
10/9 [==============================] - 60s 6s/step - loss: 1.4399 - acc: 0.9599 - val_loss: 1.5451 - val_acc: 0.5768
Epoch 36/50
10/9 [==============================] - 60s 6s/step - loss: 1.4373 - acc: 0.8998 - val_loss: 1.5424 - val_acc: 0.5768
Epoch 37/50
10/9 [==============================] - 60s 6s/step - loss: 1.4231 - acc: 0.9199 - val_loss: 1.5389 - val_acc: 0.6183
Epoch 38/50
10/9 [==============================] - 59s 6s/step - loss: 1.4247 - acc: 0.9199 - val_loss: 1.5372 - val_acc: 0.5934
Epoch 39/50
10/9 [==============================] - 60s 6s/step - loss: 1.4153 - acc: 0.9399 - val_loss: 1.5406 - val_acc: 0.5560
Epoch 40/50
10/9 [==============================] - 60s 6s/step - loss: 1.4074 - acc: 0.9800 - val_loss: 1.5327 - val_acc: 0.6224
Epoch 41/50
10/9 [==============================] - 60s 6s/step - loss: 1.4023 - acc: 0.9800 - val_loss: 1.5305 - val_acc: 0.6100
Epoch 42/50
10/9 [==============================] - 59s 6s/step - loss: 1.3938 - acc: 0.9800 - val_loss: 1.5269 - val_acc: 0.5975
Epoch 43/50
10/9 [==============================] - 60s 6s/step - loss: 1.3897 - acc: 0.9599 - val_loss: 1.5234 - val_acc: 0.6432
Epoch 44/50
10/9 [==============================] - 60s 6s/step - loss: 1.3828 - acc: 0.9800 - val_loss: 1.5210 - val_acc: 0.6556
Epoch 45/50
10/9 [==============================] - 59s 6s/step - loss: 1.3848 - acc: 0.9599 - val_loss: 1.5234 - val_acc: 0.5975
Epoch 46/50
10/9 [==============================] - 60s 6s/step - loss: 1.3716 - acc: 0.9800 - val_loss: 1.5216 - val_acc: 0.6432
Epoch 47/50
10/9 [==============================] - 60s 6s/step - loss: 1.3721 - acc: 0.9800 - val_loss: 1.5195 - val_acc: 0.6266
Epoch 48/50
10/9 [==============================] - 60s 6s/step - loss: 1.3622 - acc: 0.9599 - val_loss: 1.5108 - val_acc: 0.6141
Epoch 49/50
10/9 [==============================] - 60s 6s/step - loss: 1.3452 - acc: 0.9399 - val_loss: 1.5140 - val_acc: 0.6432
Epoch 50/50
10/9 [==============================] - 60s 6s/step - loss: 1.3387 - acc: 0.9599 - val_loss: 1.5100 - val_acc: 0.6266

from keras.models import load_model
from keras.preprocessing.image import load_img, save_img, img_to_array
from keras_vggface.utils import preprocess_input

test_img = image.load_img('test.jpg', target_size=(224, 224))
img_test = image.img_to_array(test_img)
img_test = np.expand_dims(img_test, axis=0)
img_test = utils.preprocess_input(img_test)
predictions = model.predict(img_test)
predicted_class=np.argmax(predictions,axis=1)

labels = (train_generator.class_indices)
labels = dict((v,k) for k,v in labels.items())
predictions = [labels[k] for k in predicted_class]
print(predictions)
['RobertDJr']

#Load trained model
from keras.models import load_model
from keras_vggface import utils
import cv2

image_size = 224
device_id = 0 #camera_device id 

model = load_model('my faces.h5')

#make labels according to your dataset folder 
labels = dict(fisrtname=0,secondname=1) #and so on
print(labels)

cascade_classifier = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
camera = cv2.VideoCapture(device_id)

while camera.isOpened():
    ok, cam_frame = camera.read()
    if not ok:
        break
    
    gray_img=cv2.cvtColor(cam_frame, cv2.COLOR_BGR2GRAY)
    faces= cascade_classifier.detectMultiScale(gray_img, minNeighbors=5)

    for (x,y,w,h) in faces:
        cv2.rectangle(cam_frame,(x,y),(x+w,y+h),(255,255,0),2)
        roi_color = cam_frame [y:y+h, x:x+w]
        roi color = cv2.cvtColor(roi_color, cv2.COLOR_BGR2RGB)
        roi_color = cv2.resize(roi_color, (image_size, image_size))
        image = roi_color.astype(np.float32, copy=False)
        image = np.expand_dims(image, axis=0)
        image = preprocess_input(image, version=1) # or version=2
        preds = model.predict(image)
        predicted_class=np.argmax(preds,axis=1)

        labels = dict((v,k) for k,v in labels.items())
        name = [labels[k] for k in predicted_class]

        cv2.putText(cam_frame,str(name), 
                    (x + 10, y + 10), cv2.FONT_HERSHEY_SIMPLEX, 1, (255,0,255), 2)
        
    cv2.imshow('video image', cam_frame)
    key = cv2.waitKey(30)
    if key == 27: # press 'ESC' to quit
        break

camera.release()
cv2.destroyAllWindows()