Diabetic-Retinopathy-Classification/main.py at main · MathAndMedLab/Diabetic-Retinopathy-Classification · GitHub

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
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Activation, Dropout, Flatten, Dense
from keras.applications import InceptionV3
from keras.applications import VGG16
from keras.optimizers import Adam
from keras.models import load_model

def train(test_dir, img_size, batch_size, nb_test_samples, isForTraining, isBinary, train_dir='', val_dir='',
          nb_train_samples=0, nb_validation_samples=0, number_epochs=5, saveName='', loadName=''):
    img_width, img_height = img_size, img_size
    input_shape = (img_width, img_height, 3)
    datagen = ImageDataGenerator(rescale=1. / 255)

    if isBinary:
        grades = 1
        activation_function = 'sigmoid'
        classification_type = 'binary'
    else:
        grades = 5
        activation_function = 'softmax'
        classification_type = 'categorical'

    if isForTraining:
        inception_net = InceptionV3(weights='imagenet', include_top=False, input_shape=(img_width, img_height, 3))

        inception_net.trainable = False

        model = Sequential()
        model.add(inception_net)
        model.add(Flatten())
        model.add(Dense(256))
        model.add(Activation('relu'))
        model.add(Dropout(0.5))
        model.add(Dense(grades))
        model.add(Activation(activation_function))

        model.compile(loss=classification_type + '_crossentropy',
                      optimizer=Adam(lr=1e-5),
                      metrics=['accuracy'])

        train_generator = datagen.flow_from_directory(
            train_dir,
            target_size=(img_width, img_height),
            batch_size=batch_size,
            class_mode=classification_type)

        val_generator = datagen.flow_from_directory(
            val_dir,
            target_size=(img_width, img_height),
            batch_size=batch_size,
            class_mode=classification_type)

        test_generator = datagen.flow_from_directory(
            test_dir,
            target_size=(img_width, img_height),
            batch_size=batch_size,
            class_mode=classification_type)

        model.fit(
            train_generator,
            steps_per_epoch=nb_train_samples // batch_size,
            epochs=number_epochs,
            validation_data=val_generator,
            validation_steps=nb_validation_samples // batch_size)
    else:
        model = load_model('networks/' + loadName + '.h5')

        if isBinary:
            grades += 1

        for current_grade in range(grades):
            size = 0

            for grade in range(grades):
                if grade == current_grade:
                    size = len(os.listdir(test_dir + "/" + str(grade)))
                    continue
                for img_name in os.listdir(test_dir + "/" + str(grade)):
                    os.rename(test_dir + "/" + str(grade) + "/" + img_name, str(grade) + "/" + img_name)

            test_generator = datagen.flow_from_directory(
                test_dir,
                target_size=(img_width, img_height),
                batch_size=batch_size,
                class_mode=classification_type)

            scores = model.evaluate_generator(test_generator, size // batch_size)

            print("Точность класса " + str(current_grade) + ": %.2f%%" % (scores[1]*100))

            for grade in range(grades):
                if grade == current_grade:
                    continue
                for img_name in os.listdir(str(grade)):
                    os.rename(str(grade) + "/" + img_name, test_dir + "/" + str(grade) + "/" + img_name)

        test_generator = datagen.flow_from_directory(
            test_dir,
            target_size=(img_width, img_height),
            batch_size=batch_size,
            class_mode=classification_type)

    scores = model.evaluate_generator(test_generator, nb_test_samples // batch_size)

    print(scores)
    print("Правильность на тестовых данных: %.2f%%" % (scores[1]*100))

    if isForTraining:
        model.save('networks/' + saveName + '.h5')

if __name__ == "__main__":

    i_want_to_train = True

    if i_want_to_train:
        train(train_dir='trainBalancedBin',
              val_dir='valBalancedBin',
              test_dir='testBalancedBin',
              img_size=200,
              batch_size=16,
              nb_train_samples=4064,
              nb_validation_samples=800,
              nb_test_samples=800,
              isForTraining=i_want_to_train, isBinary=True, number_epochs=2, saveName='network200')
    else:
        train(test_dir='testBalancedBin',
              img_size=200,
              batch_size=10,
              nb_test_samples=800,
              isForTraining=i_want_to_train, isBinary=True, loadName='precision200')