-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrainMobileNetV2.py
205 lines (147 loc) · 7.42 KB
/
trainMobileNetV2.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
import os
import numpy
import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt
#from tensorflow.python.client import device_lib
#print(device_lib.list_local_devices())
#tf.enable_eager_execution()
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True
train_path=r'c:\Image Recognition tenserflow\Animals_train'
save_model_name = 'train2_MobileNetV2_hiden1024.h5'
label_path = "train2_label.txt"
image_size = 224 # All images will be resized to 224x224
batch_size = 64
epochs = 20
def _process_pathnames(fname, label_path):
# We map this function onto each pathname pair
img_str = tf.read_file(fname)
img = tf.image.decode_jpeg(img_str, channels=3)
label_img_str = tf.read_file(label_path)
# These are gif images so they return as (num_frames, h, w, c)
label_img = tf.image.decode_gif(label_img_str)[0]
# The label image should only have values of 1 or 0, indicating pixel wise
# object (car) or not (background). We take the first channel only.
label_img = label_img[:, :, 0]
label_img = tf.expand_dims(label_img, axis=-1)
return img, label_img
# Reads an image from a file, decodes it into a dense tensor, and resizes it
# to a fixed shape.
def _parse_function(filename, label):
image_string = tf.io.read_file(filename)
image_decoded = tf.image.decode_jpeg(image_string)
image_resized = tf.image.resize(image_decoded, [28, 28])
return image_resized, label
classes = os.listdir(train_path)
num_classes = len(classes)
print(num_classes)
"""### Create Image Data Generator with Image Augmentation
We will use ImageDataGenerator to rescale the images.
To create the train generator, specify where the train dataset directory, image size, batch size and binary classification mode.
The validation generator is created the same way.
"""
# Rescale all images by 1./255 and apply image augmentation
train_datagen = keras.preprocessing.image.ImageDataGenerator(rescale=1./255,
validation_split=0.2,
rotation_range=5,
width_shift_range=0.2,
height_shift_range=0.2,
horizontal_flip=False,
shear_range=0.2,
zoom_range=0.2)
# Flow training images in batches of 20 using train_datagen generator
train_generator = train_datagen.flow_from_directory(
train_path, # Source directory for the training images
target_size=(image_size, image_size),
batch_size=batch_size,
class_mode='categorical',
subset='training')
# Flow validation images in batches of 20 using test_datagen generator
validation_generator = train_datagen.flow_from_directory(
train_path, # Source directory for the validation images
target_size=(image_size, image_size),
batch_size=batch_size,
class_mode='categorical',
subset='validation')
with open(label_path, "w") as txt_file:
for cls in train_generator.class_indices:
txt_file.write(cls + "\n") # works with any number of elements in a line
"""## Create the base model from the pre-trained convnets
We will create the base model from the **MobileNet V2** model developed at Google, and pre-trained on the ImageNet dataset, a large dataset of 1.4M images and 1000 classes of web images. This is a powerful model. Let's see what the features that it has learned can do for our cat vs. dog problem.
First, we need to pick which intermediate layer of MobileNet V2 we will use for feature extraction. A common practice is to use the output of the very last layer before the flatten operation, the so-called "bottleneck layer". The reasoning here is that the following fully-connected layers will be too specialized to the task the network was trained on, and thus the features learned by these layers won't be very useful for a new task. The bottleneck features, however, retain much generality.
Let's instantiate an MobileNet V2 model pre-loaded with weights trained on ImageNet. By specifying the **include_top=False** argument, we load a network that doesn't include the classification layers at the top, which is ideal for feature extraction.
"""
IMG_SHAPE = (image_size, image_size, 3)
# Create the base model from the pre-trained model MobileNet V2
base_model = tf.keras.applications.MobileNetV2(input_shape=IMG_SHAPE,
include_top=False,
weights='imagenet')
"""## Feature extraction
We will freeze the convolutional base created from the previous step and use that as a feature extractor, add a classifier on top of it and train the top-level classifier.
### Freeze the convolutional base
It's important to freeze the convolutional based before we compile and train the model. By freezing (or setting `layer.trainable = False`), we prevent the weights in these layers from being updated during training.
"""
base_model.trainable = False
# Let's take a look at the base model architecture
#base_model.summary()
"""#### Add a classification head
Now let's add a few layers on top of the base model:
"""
model = tf.keras.Sequential([
base_model,
keras.layers.GlobalAveragePooling2D(),
#keras.layers.Dense(1, activation='sigmoid')
#keras.layers.Dense(units=train_generator.num_classes, activation=tf.nn.relu),
keras.layers.Dense(units=1024, activation=tf.nn.relu),
keras.layers.Dense(units=train_generator.num_classes, activation=tf.nn.softmax)
])
"""### Compile the model
You must compile the model before training it.
"""
model.compile(optimizer=tf.keras.optimizers.RMSprop(lr=0.0001),
loss='categorical_crossentropy',
metrics=['accuracy'])
model.summary()
"""These 1.2K trainable parameters are divided among 2 TensorFlow `Variable` objects, the weights and biases of the two dense layers:"""
len(model.trainable_variables)
"""### Train the model
After training for 10 epochs, we are able to get ~94% accuracy.
If you have more time, train it to convergence (50 epochs, ~96% accuracy)
"""
steps_per_epoch = train_generator.n // batch_size
validation_steps = validation_generator.n // batch_size
history = model.fit_generator(train_generator,
steps_per_epoch = steps_per_epoch,
epochs=epochs,
workers=4,
validation_data=validation_generator,
validation_steps=validation_steps)
# save model and architecture to single file
model.save(save_model_name)
print("Saved model to disk")
"""### Learning curves
Let's take a look at the learning curves of the training and validation accuracy / loss, when using the MobileNet V2 base model as a fixed feature extractor.
If you train to convergence (`epochs=50`) the resulting graph should look like this:
"""
acc = history.history['acc']
val_acc = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']
plt.figure(figsize=(8, 8))
plt.subplot(2, 1, 1)
plt.plot(acc, label='Training Accuracy')
plt.plot(val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.ylabel('Accuracy')
plt.ylim([min(plt.ylim()),1])
plt.title('Training and Validation Accuracy')
plt.subplot(2, 1, 2)
plt.plot(loss, label='Training Loss')
plt.plot(val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.ylabel('Cross Entropy')
plt.ylim([0,max(plt.ylim())])
plt.title('Training and Validation Loss')
plt.show()