-
Notifications
You must be signed in to change notification settings - Fork 0
/
evaluate_classification_models.py
188 lines (154 loc) · 7.58 KB
/
evaluate_classification_models.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
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
180
181
182
183
184
185
186
187
188
import matplotlib.pyplot as plt
import numpy as np
import matplotlib.pyplot as plt
import torch
from data import CrowdClassificationDataSet
from data import default_train_transform_classification, default_val_transform_classification
import argparse
from sklearn.metrics import accuracy_score, confusion_matrix, ConfusionMatrixDisplay, f1_score, precision_score, recall_score
def run_argparse():
parser = argparse.ArgumentParser()
parser.add_argument("--model", type=str)
return parser.parse_args()
def main(args):
loaders = {
"train": CrowdClassificationDataSet(
'part_A/train_data', default_train_transform_classification()
),
"val": CrowdClassificationDataSet(
'part_A/test_data', default_val_transform_classification()
),
"test_unbalanced": CrowdClassificationDataSet(
'part_B/test_data', default_val_transform_classification()
)
}
model = torch.load('saved_models/' + args.model)
model.eval()
train_vgg16_predictions = []
train_vgg16_actual = []
val_vgg16_predictions = []
val_vgg16_actual = []
test_b_vgg16_predictions = []
test_b_vgg16_actual = []
test_ub_vgg16_predictions = []
test_ub_vgg16_actual = []
print('Evaluating Training...')
for i, data in enumerate(loaders['train'], 0):
dt = data
image = dt['image'].to()
bin = dt['bin']
model.eval()
outputs = model(image[None, ...].float())
expected = torch.Tensor([bin]).type(torch.LongTensor)
_, preds = torch.max(outputs, 1)
train_vgg16_predictions.append(preds)
train_vgg16_actual.append(expected)
limit = int(len(loaders['val']) / 2)
print('Evaluating Validation...')
for i, data in enumerate(loaders['val'], 0):
if i >= limit: break
dt = data
image = dt['image'].to()
bin = dt['bin']
model.eval()
outputs = model(image[None, ...].float())
expected = torch.Tensor([bin]).type(torch.LongTensor)
_, preds = torch.max(outputs, 1)
val_vgg16_predictions.append(preds)
val_vgg16_actual.append(expected)
print('Evaluating Testing Balanced...')
for i, data in enumerate(loaders['val'], limit):
dt = data
image = dt['image'].to()
bin = dt['bin']
model.eval()
outputs = model(image[None, ...].float())
expected = torch.Tensor([bin]).type(torch.LongTensor)
_, preds = torch.max(outputs, 1)
test_b_vgg16_predictions.append(preds)
test_b_vgg16_actual.append(expected)
print('Evaluating Testing Unbalanced...')
for i, data in enumerate(loaders['test_unbalanced'], 0):
dt = data
image = dt['image'].to()
bin = dt['bin']
model.eval()
outputs = model(image[None, ...].float())
expected = torch.Tensor([bin]).type(torch.LongTensor)
_, preds = torch.max(outputs, 1)
test_ub_vgg16_predictions.append(preds)
test_ub_vgg16_actual.append(expected)
train_acc = accuracy_score(train_vgg16_actual, train_vgg16_predictions)
val_acc = accuracy_score(val_vgg16_actual, val_vgg16_predictions)
test_b_acc = accuracy_score(test_b_vgg16_actual, test_b_vgg16_predictions)
test_ub_acc = accuracy_score(test_ub_vgg16_actual, test_ub_vgg16_predictions)
train_f1 = f1_score(train_vgg16_actual, train_vgg16_predictions, average='weighted')
val_f1 = f1_score(val_vgg16_actual, val_vgg16_predictions, average='weighted')
test_b_f1 = f1_score(test_b_vgg16_actual, test_b_vgg16_predictions, average='weighted')
test_ub_f1 = f1_score(test_ub_vgg16_actual, test_ub_vgg16_predictions, average='weighted')
train_prec = precision_score(train_vgg16_actual, train_vgg16_predictions, average='weighted')
val_prec = precision_score(val_vgg16_actual, val_vgg16_predictions, average='weighted')
test_b_prec = precision_score(test_b_vgg16_actual, test_b_vgg16_predictions, average='weighted')
test_ub_prec = precision_score(test_ub_vgg16_actual, test_ub_vgg16_predictions, average='weighted')
train_recall = recall_score(train_vgg16_actual, train_vgg16_predictions, average='weighted')
val_recall = recall_score(val_vgg16_actual, val_vgg16_predictions, average='weighted')
test_b_recall = recall_score(test_b_vgg16_actual, test_b_vgg16_predictions, average='weighted')
test_ub_recall = recall_score(test_ub_vgg16_actual, test_ub_vgg16_predictions, average='weighted')
print("{}".format(args.model))
print('================================')
print('Training Accuracy: {}'.format(train_acc))
print('Validation Accuracy: {}'.format(val_acc))
print('Testing (Balanced) Accuracy: {}'.format(test_b_acc))
print('Testing (Unbalanced) Accuracy: {}'.format(test_ub_acc))
print('================================')
print('Training f1: {}'.format(train_acc))
print('Validation f1: {}'.format(val_acc))
print('Testing (Balanced) f1: {}'.format(test_b_acc))
print('Testing (Unbalanced) f1: {}'.format(test_ub_acc))
print('================================')
print('Training Precision: {}'.format(train_prec))
print('Validation Precision: {}'.format(val_prec))
print('Testing (Balanced) Precision: {}'.format(test_b_prec))
print('Testing (Unbalanced) Precision: {}'.format(test_ub_prec))
print('================================')
print('Training Recall: {}'.format(train_recall))
print('Validation Recall: {}'.format(val_recall))
print('Testing (Balanced) Recall: {}'.format(test_b_recall))
print('Testing (Unbalanced) Recall: {}'.format(test_ub_recall))
print('================================')
fg, (p1, p2, p3, p4) = plt.subplots(1, 4, figsize=(15, 4))
cf_matrix = confusion_matrix(train_vgg16_actual, train_vgg16_predictions, labels=[0, 1, 2, 3, 4])
disp = ConfusionMatrixDisplay(cf_matrix, display_labels=[0, 1, 2, 3, 4])
disp.plot(ax=p1)
disp.ax_.set_title('Training Acc: {:.2f}\n Prec: {:.2f}'.format(train_acc, train_prec))
disp.im_.colorbar.remove()
disp.ax_.set_xlabel('')
cf_matrix = confusion_matrix(val_vgg16_actual, val_vgg16_predictions, labels=[0, 1, 2, 3, 4])
disp = ConfusionMatrixDisplay(cf_matrix, display_labels=[0, 1, 2, 3, 4])
disp.plot(ax=p2)
disp.ax_.set_title('Validation Acc: {:.2f}\n Prec: {:.2f}'.format(val_acc, val_prec))
disp.im_.colorbar.remove()
disp.ax_.set_xlabel('')
disp.ax_.set_ylabel('')
cf_matrix = confusion_matrix(test_b_vgg16_actual, test_b_vgg16_predictions, labels=[0, 1, 2, 3, 4])
disp = ConfusionMatrixDisplay(cf_matrix, display_labels=[0, 1, 2, 3, 4])
disp.plot(ax=p3)
disp.ax_.set_title('Testing (Balanced) Acc: {:.2f}\n Prec: {:.2f}'.format(test_b_acc, test_b_prec))
disp.im_.colorbar.remove()
disp.ax_.set_xlabel('')
disp.ax_.set_ylabel('')
cf_matrix = confusion_matrix(test_ub_vgg16_actual, test_ub_vgg16_predictions, labels=[0, 1, 2, 3, 4])
disp = ConfusionMatrixDisplay(cf_matrix, display_labels=[0, 1, 2, 3, 4])
disp.plot(ax=p4)
disp.ax_.set_title('Testing (Unbalanced) Acc: {:.2f}\n Prec: {:.2f}'.format(test_ub_acc, test_ub_prec))
disp.im_.colorbar.remove()
disp.ax_.set_xlabel('')
disp.ax_.set_ylabel('')
fg.text(0.04, 0.5, 'Actual Label', va='center', rotation='vertical')
fg.text(0.5, 0.04, 'Predicted Label', ha='center')
plt.subplots_adjust(wspace=0.40, hspace=0.1)
fg.colorbar(disp.im_, ax=(p1, p2, p3, p4))
fg.savefig('results/{}_results'.format(args.model))
if __name__=='__main__':
args = run_argparse()
main(args)