-
Notifications
You must be signed in to change notification settings - Fork 7
/
result.py
96 lines (70 loc) · 2.55 KB
/
result.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
import os
import numpy as np
import matplotlib.pyplot as plt
import torch
def class_level_accuracy(model, loader, device, classes):
"""Print test accuracy for each class in dataset.
Args:
model: Model instance.
loader: Data loader.
device: Device where data will be loaded.
classes: List of classes in the dataset.
"""
class_correct = list(0. for i in range(len(classes)))
class_total = list(0. for i in range(len(classes)))
with torch.no_grad():
for _, (images, labels) in enumerate(loader, 0):
images, labels = images.to(device), labels.to(device)
outputs = model(images)
_, predicted = torch.max(outputs, 1)
c = (predicted == labels).squeeze()
for i in range(len(labels)):
label = labels[i]
class_correct[label] += c[i].item()
class_total[label] += 1
for i in range(10):
print('Accuracy of %5s : %2d %%' % (classes[i], 100 * class_correct[i] / class_total[i]))
def plot_metric(values, metric):
# Initialize a figure
fig = plt.figure(figsize=(7, 5))
# Plot values
plt.plot(values)
# Set plot title
plt.title(f'Validation {metric}')
# Label axes
plt.xlabel('Epoch')
plt.ylabel(metric)
# Set legend
location = 'upper' if metric == 'Loss' else 'lower'
# Save plot
fig.savefig(f'{metric.lower()}_change.png')
def save_and_show_result(data, classes):
"""Display 25 misclassified images.
Args:
data: Contains model predictions and labels.
"""
# Create directories for saving data
path = os.path.join(
os.path.dirname(os.path.abspath(__file__)), 'predictions'
)
if not os.path.exists(path):
os.makedirs(path)
# Initialize plot
row_count = -1
fig, axs = plt.subplots(5, 5, figsize=(10, 10))
fig.tight_layout()
for idx, result in enumerate(data):
# If 25 samples have been stored, break out of loop
if idx > 24:
break
rgb_image = np.transpose(result['image'], (1, 2, 0)) / 2 + 0.5
label = result['label'].item()
prediction = result['prediction'].item()
# Plot image
if idx % 5 == 0:
row_count += 1
axs[row_count][idx % 5].axis('off')
axs[row_count][idx % 5].set_title(f'Label: {classes[label]}\nPrediction: {classes[prediction]}')
axs[row_count][idx % 5].imshow(rgb_image)
# Save image
fig.savefig(f'{path}/incorrect_predictions.png', bbox_inches='tight')