AutoAIF/tests.py at main · petmri/AutoAIF · 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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
import unittest
import numpy as np
import nibabel as nib
import os
import tensorflow as tf
from utils_vif import *
from model_vif import *

class TestDataGenerator(unittest.TestCase):

    def setUp(self):
        DATASET_DIR = "/home/mrispec/AUTOAIF_DATA/loos_model"
        self.paths = ['203421_1st_timepoint.nii.gz', '500101_1st_timepoint.nii.gz', '1101970_1st_timepoint.nii.gz']
        # self.paths = ['sub-1100841_ses-02_desc-hmc_DCE.nii.gz']
        self.directory = '/home/mrispec/AUTOAIF_DATA/loos_model/USC/'
        self.batch_size = 1
        self.input_size = (256, 256, 32, 32)
        self.shuffle = True
        self.data_augmentation = True
        # self.currentgenerator = DataGenerator(self.paths, self.directory, self.batch_size, self.input_size, self.shuffle, self.data_augmentation)
        # train_set = load_data(os.path.join(self.directory, 'images/', self.paths[0]))
        # self.tfgenerator = tf.data.Dataset.from_generator(lambda: train_generator(os.path.join(DATASET_DIR,"train/"), True, True, self.paths), output_types=(tf.float32, (tf.float32, tf.float32, tf.float32))).batch(256).prefetch(tf.data.AUTOTUNE)
        # self.dataset = tf.data.Dataset.from_generator(generator=self.generator,
        #                                      output_types=(tf.float32, tf.float32),
        #                                      output_shapes=((self.batch_size, *self.input_size), (self.batch_size, 1)))
        self.num_epochs = 2

    # def test_len(self):
    #     self.assertEqual(len(self.currentgenerator), len(self.paths) // self.batch_size)

    # def test_getitem(self):
    #     batch_images, (batch_cof, batch_curve, batch_vol, batch_quality) = self.currentgenerator[0]
    #     self.assertEqual(batch_images.shape, (self.batch_size, *self.input_size))
    #     # self.assertEqual(batch_cof.shape, (self.batch_size, 3))
    #     self.assertEqual(batch_curve.shape, (self.batch_size, self.input_size[3]))
    #     self.assertEqual(batch_vol.shape, (self.batch_size, 1))
    #     self.assertEqual(batch_quality.shape, (self.batch_size, 1))

    # # def test_on_epoch_end(self):
    # #     self.generator.on_epoch_end()
    # #     self.assertFalse(np.array_equal(self.generator.indexes, np.arange(len(self.paths))))

    # def test_data_generation(self):
    #     list_IDs_temp = self.paths[:self.batch_size]
    #     batch_images, (batch_masks, batch_curve, batch_vol, batch_quality) = self.currentgenerator._DataGenerator__data_generation(list_IDs_temp)
    #     self.assertEqual(batch_images.shape, (self.batch_size, *self.input_size))
    #     self.assertEqual(batch_masks.shape, (self.batch_size, self.input_size[0], self.input_size[1], self.input_size[2], 1))
    #     self.assertEqual(batch_curve.shape, (self.batch_size, self.input_size[3]))
    #     self.assertEqual(batch_vol.shape, (self.batch_size, 1))
    #     self.assertEqual(batch_quality.shape, (self.batch_size, 1))
    #     #print a batch of curves
    #     print(batch_curve)
    #     print(batch_vol)

    def test_TFRecord(self):
        imgs = [os.path.join(self.directory, 'images/', img) for img in self.paths]
        masks = [os.path.join(self.directory, 'masks/', mask) for mask in self.paths]
        print(imgs)
        img = self.paths[0]
        print(img)
        img = os.path.join(self.directory, 'images/', img)
        mask = os.path.join(self.directory, 'masks/', img)
        write_records(imgs, masks, len(self.paths), './test/train')

        list_of_records=['./test/train_000-of-000.tfrecords']
        batch_size=1
        ds = get_batched_dataset(list_of_records, batch_size=batch_size, shuffle_size=1)
        image, label = next(iter(ds))
        _ = plt.imshow(image[0, :, :, 0, 0], cmap='gray')
        plt.show()
        # _ = plt.title(label)
        def visualize(original, augmented):
            fig = plt.figure()
            plt.subplot(1,2,1)
            plt.title('Original image')
            plt.imshow(original[0, :, :, 0, 0], cmap='gray')

            plt.subplot(1,2,2)
            plt.title('Augmented image')
            plt.imshow(augmented[0, :, :, 0, 0], cmap='gray')

        # flipped = tf.image.flip_left_right(image[0, :, :, :, 0])
        # visualize(image, flipped)


        # (Xs, Ys) = next(ds.as_numpy_iterator())

        # (batch_size, )
        # order will depend on shuffle (turn off all shuffling to verify order)
        # print(Ys.shape)

        # (batch_size, x_dim, y_dim, z_dim, 1)
        # print(Xs.shape)
        model = unet3d( img_size        = (X_DIM, Y_DIM, Z_DIM, T_DIM),
                learning_rate   = 1e-3,
                learning_decay  = 1e-9,
                weights         = [0, 1, 0])

        model.fit(ds, validation_data=ds, epochs=1, steps_per_epoch=len(self.paths)//batch_size, validation_steps=len(self.paths)//batch_size)

    # def test_TensorRT(self):
    #     import tensorflow as tf
    #     from tensorflow.python.compiler.tensorrt import trt_convert as trt

    #     list_of_records=['train_000-of-000.tfrecords']
    #     train = get_batched_dataset(list_of_records, batch_size=self.batch_size, shuffle_size=1)
    #     list_of_records=['val_000-of-000.tfrecords']
    #     val = get_batched_dataset(list_of_records, batch_size=self.batch_size, shuffle_size=1)
    #     model = unet3d( img_size        = (X_DIM, Y_DIM, Z_DIM, T_DIM),
    #             learning_rate   = 1e-3,
    #             learning_decay  = 1e-9,
    #             weights         = [0, 1, 0])

    #     model.fit(train, validation_data=val, epochs=1, steps_per_epoch=len(self.paths)//self.batch_size, validation_steps=len(self.paths)//self.batch_size)
    #     model.save('/home/mrispec/AUTOAIF_DATA/weights/trttest/model.keras')

    #     # Load the model
    #     model = tf.keras.models.load_model('/home/mrispec/AUTOAIF_DATA/weights/trttest/model.keras', safe_mode=False)

    #     converter = trt.TrtGraphConverterV2(input_saved_model_dir=model)

    #     converter.convert()

    #     converter.save('/home/mrispec/AUTOAIF_DATA/weights/trttest/model_weight_TRT')

    #     trt_model = tf.keras.models.load_model('/home/mrispec/AUTOAIF_DATA/weights/trttest/model_weight_TRT')


    #     print('Loading data')
    #     volume_img = nib.load('/home/mrispec/AUTOAIF_DATA/loos_model/train/203421_1st_timepoint.nii')
    #     print(volume_img.shape)
    #     volume_data = volume_img.get_fdata()

    #     print('Preprocessing')
    #     vol_pre = preprocessing(volume_data)

    #     trt_model.predict(vol_pre)

        # print('Loading model')
        # model = unet3d(img_size = (X_DIM, Y_DIM, Z_DIM, T_DIM),\
        #                 learning_rate = 1e-3,\
        #                 learning_decay = 1e-9)
        # model.trainable = False
        # model.load_weights(args.model_weight_path)

        # print('Prediction')
        # y_pred_mask, y_pred_vf, _ = model.predict('/home/mrispec/AUTOAIF_DATA/weights/newtftest/model_weight.h5')


if __name__ == '__main__':
    unittest.main()