-
Notifications
You must be signed in to change notification settings - Fork 9
/
train.py
75 lines (63 loc) · 2.41 KB
/
train.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
import os
from keras.layers import Dense, Dropout, Flatten, MaxPooling2D, Convolution2D
from keras.models import Sequential
from keras.preprocessing.image import ImageDataGenerator
import keras
from keras.optimizers import Adam
NUM_CHANNELS = 3
IMAGE_WIDTH = 224 # Original: 455
IMAGE_HEIGHT = 224 # Original: 256
NUM_CLASSES = 3
base_script_name = os.path.splitext(__file__)[0]
filepath=base_script_name + "-{epoch:02d}-val_acc-{val_acc:.2f}.hdf5"
def get_generator(directory, train):
if train:
datagen = ImageDataGenerator(
rescale=1./255,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True)
else:
datagen = ImageDataGenerator(rescale=1./255)
return datagen.flow_from_directory(
directory=directory,
target_size=(IMAGE_WIDTH, IMAGE_HEIGHT),
batch_size=8,
class_mode='categorical')
model = Sequential([
Convolution2D(16, 3, 3, border_mode='same', subsample=(2, 2), input_shape=(IMAGE_WIDTH, IMAGE_HEIGHT, NUM_CLASSES), activation='relu'),
MaxPooling2D(pool_size=(3, 3)),
Dropout(0.2),
Convolution2D(32, 3, 3, border_mode='same', activation='relu'),
MaxPooling2D(pool_size=(3, 3)),
Dropout(0.2),
Convolution2D(64, 3, 3, border_mode='same', activation='relu'),
MaxPooling2D(pool_size=(2, 2)),
Dropout(0.2),
Flatten(),
Dense(128, activation='tanh'),
Dropout(0.3),
Dense(NUM_CLASSES, activation='softmax'),
])
model.summary()
directory = './data/'
train_generator = get_generator(directory+'./train', True)
validation_generator = get_generator(directory+'valid', False)
# we need to recompile the model for these modifications to take effect
# we use SGD with a low learning rate
model.compile(optimizer=Adam(lr=0.0003), loss='categorical_crossentropy', metrics=['accuracy'])
# Callbacks
checkpoint = keras.callbacks.ModelCheckpoint(filepath, monitor='val_acc', verbose=1, save_best_only=False, mode='auto')
tensorboard = keras.callbacks.TensorBoard(log_dir='./tensorboar', histogram_freq=0, write_graph=True, write_images=True)
callbacks = [checkpoint, tensorboard]
# we train our model again (this time fine-tuning the top 2 inception blocks
# alongside the top Dense layers
model.fit_generator(
train_generator,
samples_per_epoch=train_generator.N,
nb_epoch=200,
validation_data=validation_generator,
nb_val_samples=validation_generator.N,
callbacks=callbacks,
)
model.evaluate_generator(validation_generator, val_samples=len(validation_generator.filenames))