-
Notifications
You must be signed in to change notification settings - Fork 1
/
train_personalized.py
145 lines (111 loc) · 4.6 KB
/
train_personalized.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
from distutils import config
import math
import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.layers import Dense, Flatten, Conv2D, MaxPooling2D, Dropout, LeakyReLU, GlobalAveragePooling2D
from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import SGD, Adam
from tensorflow.keras.utils import plot_model
import numpy as np
import matplotlib.pyplot as plt
from sklearn.metrics import classification_report
from tensorflow.keras.callbacks import ReduceLROnPlateau
import datetime
# Caminho para o diretório com as imagens processadas
dataset_dir = "Letras"
dataset_validation_dir = "test"
FILE_NAME="cnn_model_libras"
def getDateStr():
return str('{date:%Y%m%d_%H%M}').format(date=datetime.datetime.now())
# Parâmetros de pré-processamento
image_size = (64, 64) # Tamanho das imagens para o treinamento
batch_size = 64 # Número de amostras por lote
# Preparando os dados
train_datagen = ImageDataGenerator(
rescale=1./255, # Normalização das imagens,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
validation_split=0.25 # 25% dos dados para validação
)
train_generator = train_datagen.flow_from_directory(
dataset_dir,
target_size=image_size,
batch_size=batch_size,
class_mode='categorical', # Modo "categorical" para classificação multiclasse,
shuffle=False,
)
validation_generator = train_datagen.flow_from_directory(
dataset_validation_dir,
target_size=image_size,
batch_size=batch_size,
class_mode='categorical',
shuffle=False,
)
model_constructors = {
"DenseNet201": tf.keras.applications.DenseNet201,
}
base_model = model_constructors["DenseNet201"](weights='imagenet', include_top=False, input_shape=(64,64,3))
for layer in base_model.layers:
layer.trainable=False
print(len(train_generator.class_indices))
def build(width, height, channels, classes):
inputShape = (height, width, channels)
model = Sequential()
model.add(Conv2D(filters = 32, kernel_size = (3,3), padding = 'same', input_shape = inputShape))
model.add(LeakyReLU(alpha=0.1))
model.add(MaxPooling2D((2,2)))
model.add(Conv2D(filters = 32, kernel_size = (3,3)))
model.add(LeakyReLU(alpha=0.1))
model.add(MaxPooling2D((2,2)))
model.add(Conv2D(filters = 64, kernel_size = (3,3)))
model.add(LeakyReLU(alpha=0.1))
model.add(MaxPooling2D((2,2)))
model.add(Flatten())
model.add(Dense(254, activation = 'relu'))
model.add(Dropout(0.5))
model.add(Dense(classes, activation = 'softmax'))
return model
# Compilando o modelo
#model.compile(optimizer=Adam(lr=0.001), loss='categorical_crossentropy', metrics=['accuracy'])
model = build(64, 64, 3, len(train_generator.class_indices))
early = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=15, mode='min', verbose=1)
model.compile(optimizer=SGD(0.01),
loss="categorical_crossentropy",
metrics=['accuracy'])
# Treinamento do modelo
classifier = model.fit(
train_generator,
steps_per_epoch=train_generator.samples // batch_size,
validation_data=validation_generator,
validation_steps=validation_generator.samples // batch_size,
shuffle = False,
verbose=2,
epochs=50,
callbacks=[early]
)
# Salvar o modelo treinado
print("[INFO] Salvando modelo treinado ...")
file_date = getDateStr()
model.save('../models/'+FILE_NAME+file_date+'.h5')
EPOCHS = len(classifier.history["loss"])
print('[INFO] Summary: ')
model.summary()
print("\n[INFO] Avaliando a CNN...")
score = model.evaluate_generator(generator=validation_generator, steps=(train_generator.n // validation_generator.batch_size), verbose=1)
print('[INFO] Accuracy: %.2f%%' % (score[1]*100), '| Loss: %.5f' % (score[0]))
print("[INFO] Sumarizando loss e accuracy para os datasets 'train' e 'test'")
plt.style.use("ggplot")
plt.figure()
plt.plot(np.arange(0,EPOCHS), classifier.history["loss"], label="train_loss")
plt.plot(np.arange(0,EPOCHS), classifier.history["val_loss"], label="val_loss")
plt.plot(np.arange(0,EPOCHS), classifier.history["acc"], label="train_acc")
plt.plot(np.arange(0,EPOCHS), classifier.history["val_acc"], label="val_acc")
plt.title("Training Loss and Accuracy")
plt.xlabel("Epoch #")
plt.ylabel("Loss/Accuracy")
plt.legend()
plt.savefig('../models/graphics/'+FILE_NAME+file_date+'.png', bbox_inches='tight')
print('[INFO] Gerando imagem do modelo de camadas da CNN')
plot_model(model, to_file='../models/image/'+FILE_NAME+file_date+'.png', show_shapes = True)
print('\n\n')