SCAI-Net/scai_test.py at main · Deforestation-Detector/SCAI-Net · 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
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
import scai_utils as su
import pandas as pd
import unittest
from sklearn.preprocessing import MultiLabelBinarizer
import tensorflow as tf
import numpy as np

class SCAITestMethods(unittest.TestCase):
    def test_setNLabels(self):
        IMAGE_NAMES = ['image1', 'image2', 'image3', 'image4', 'image5', 'image6']

        d_empty = {'image_name': [], 'tags': []}
        expected_num_labels_unique = 0
        df_empty = pd.DataFrame(data=d_empty)
        su.set_NLABELS(df_empty)
        num_labels_observed = su.N_LABELS
        self.assertEqual(num_labels_observed, expected_num_labels_unique)
        su.N_LABELS = None

        d_non_unique_labels = {'image_name': IMAGE_NAMES, 'tags': ['label1 label2', 'label2 label1', 'label3 label5', 'label1 label2', 'label4 label3', 'label5 label1']}
        df_non_unique_labels = pd.DataFrame(data=d_non_unique_labels)
        expected_num_labels_non_unique = 5
        su.set_NLABELS(df_non_unique_labels)
        num_labels_observed = su.N_LABELS
        self.assertEqual(num_labels_observed, expected_num_labels_non_unique)
        su.N_LABELS = None

        d_unique_labels = {'image_name': IMAGE_NAMES, 'tags': ['label1', 'label2', 'label3', 'label4', 'label5', 'label6']}
        expected_num_labels_unique = 6
        df_unique_labels = pd.DataFrame(data=d_unique_labels)
        su.set_NLABELS(df_unique_labels)
        num_labels_observed = su.N_LABELS
        self.assertEqual(num_labels_observed, expected_num_labels_unique)
        su.N_LABELS = None

        d_none = {'image_name': IMAGE_NAMES, 'tags': [None, None, None, None, None, None]}
        expected_num_labels_unique = None
        dfnone = pd.DataFrame(data=d_none)
        su.set_NLABELS(dfnone)
        num_labels_observed = su.N_LABELS
        self.assertEqual(num_labels_observed, expected_num_labels_unique)
        su.N_LABELS = None

        d_improper_column_names = {'other_images': IMAGE_NAMES, 'other_tags': ['label1 label2', 'label2 label1', 'label3 label5', 'label1 label2', 'label4 label3', 'label5 label1']}
        expected_num_labels_improper_column_names = None
        df_improper_column_names = pd.DataFrame(data=d_improper_column_names)
        su.set_NLABELS(df_improper_column_names)
        num_labels_observed = su.N_LABELS
        self.assertEqual(num_labels_observed, expected_num_labels_improper_column_names)
        su.N_LABELS = None

    def test_createData(self):
        train_dataframe = pd.read_csv('data/MINE/MINE.csv').astype(str)
        train_dataframe['image_name'] += '.jpg'
        su.set_NLABELS(train_dataframe)

        mlb = MultiLabelBinarizer()
        mlb.fit(train_dataframe["tags"].str.split(" "))
        classes = mlb.classes_

        multihot_df = pd.DataFrame(
            mlb.fit_transform(
                train_dataframe['tags'].str.split(' ')),
            columns=classes)

        train_df = pd.concat([train_dataframe[['image_name']], multihot_df], axis=1)
        train_dataframe = None

        train_dg, val_dg = su.create_data(train_df, classes)
        train_dg_is_none = train_dg == None
        val_dg_is_none = val_dg == None

        self.assertEqual(train_dg_is_none, False)
        self.assertEqual(val_dg_is_none, False)

        train_df_none = None
        train_dg_none, val_dg_none = su.create_data(train_df_none, classes)
        self.assertEqual(train_dg_none, None)
        self.assertEqual(val_dg_none, None)

        classes_none = None
        train_dg_none, val_dg_none = su.create_data(train_df, classes_none)
        self.assertEqual(train_dg_none, None)
        self.assertEqual(val_dg_none, None)

        train_df_none, classes_none = None, None
        train_dg_none, val_dg_none = su.create_data(train_df_none, classes_none)
        self.assertEqual(train_dg_none, None)
        self.assertEqual(val_dg_none, None)

    def test_f1(self):
        y_true_equal = tf.constant([0.8, 0.2, 0.8])
        y_pred_equal = tf.constant([0.4, 0.1, 0.7])
        f1_score_expected = np.around(np.float32(0.571), 3)
        f1_score_equal = np.around(su.f1(y_true_equal, y_pred_equal), 3)
        self.assertEqual(f1_score_expected, f1_score_equal)

        y_true_inequal = tf.constant([0.9, 0.8, 0.4])
        y_pred_inequal = tf.constant([0.1, 0.1, 0.6])
        f1_score_inequal = np.around(su.f1(y_true_inequal, y_pred_inequal), 3)
        f1_score_expected = np.around(np.float32(0.258), 3)
        self.assertEqual(f1_score_inequal, f1_score_expected)

        y_true_invalid = [0.8, 0.2, 0.8]
        y_pred_invalid = [0.8, 0.2, 0.8]
        f1_score_expected = None
        f1_score_invalid = su.f1(y_true_invalid, y_pred_invalid)
        self.assertEqual(f1_score_expected, f1_score_invalid)

    def test_f1Loss(self):
        y_true_equal = tf.constant([0.8, 0.2, 0.8])
        y_pred_equal = tf.constant([0.4, 0.1, 0.7])
        f1Loss_score_expected = np.around(np.float32(0.4), 3)
        f1Loss_score_equal = np.around(su.f1_loss(y_true_equal, y_pred_equal), 3)
        self.assertEqual(f1Loss_score_expected, f1Loss_score_equal)

        y_true_inequal = tf.constant([0.9, 0.8, 0.4])
        y_pred_inequal = tf.constant([0.1, 0.1, 0.6])
        f1Loss_score_inequal = np.around(su.f1_loss(y_true_inequal, y_pred_inequal), 3)
        f1Loss_score_expected = np.around(np.float32(0.717), 3)
        self.assertEqual(f1Loss_score_inequal, f1Loss_score_expected)

        y_true_invalid = [0.8, 0.2, 0.8]
        y_pred_invalid = [0.8, 0.2, 0.8]
        f1Loss_score_expected = None
        f1Loss_score_invalid = su.f1_loss(y_true_invalid, y_pred_invalid)
        self.assertEqual(f1Loss_score_expected, f1Loss_score_invalid)

    def test_createTransferModel(self):
        su.N_LABELS = 1

        ARCH_IN_ARCHITECTURES = 'MobileNetV2'
        expected_transfer_model = su.create_transfer_model(ARCH_IN_ARCHITECTURES)
        self.assertNotEqual(expected_transfer_model, None)

        ARCH_NOT_IN_ARCHITECTURES = 'MobileNetV3'
        expected_transfer_model = su.create_transfer_model(ARCH_NOT_IN_ARCHITECTURES)
        self.assertEqual(expected_transfer_model, None)

        INVALID_ARCH = None
        expected_transfer_model = su.create_transfer_model(INVALID_ARCH)
        self.assertEqual(expected_transfer_model, None)

        su.N_LABELS = None

    def test_reverseHot(self):
        label_valid = np.array([0, 2])
        classes_valid = ['label1', 'label2', 'label3']
        reverse_hot_valid = su.reverseHot(label_valid, classes_valid)
        reverse_hot_expected = 'label1 label3'
        self.assertEqual(reverse_hot_valid, reverse_hot_expected)

        label_invalid = [0, 2]
        classes_valid = ['label1', 'label2', 'label3']
        reverse_hot_invalid = su.reverseHot(label_invalid, classes_valid)
        reverse_hot_expected = None
        self.assertEqual(reverse_hot_invalid, reverse_hot_expected)

        label_valid = np.array([0, 2])
        classes_invalid = 'label'
        reverse_hot_invalid = su.reverseHot(label_valid, classes_invalid)
        reverse_hot_expected = None
        self.assertEqual(reverse_hot_invalid, reverse_hot_expected)

        label_valid = np.array([0, 5])
        classes_invalid = ['label1', 'label2', 'label3']
        reverse_hot_invalid = su.reverseHot(label_valid, classes_invalid)
        reverse_hot_expected = None
        self.assertEqual(reverse_hot_invalid, reverse_hot_expected)

        label_valid = np.array([0, 1])
        classes_invalid = ['label1', 1, 'label3']
        reverse_hot_invalid = su.reverseHot(label_valid, classes_invalid)
        reverse_hot_expected = None
        self.assertEqual(reverse_hot_invalid, reverse_hot_expected)


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