-
Notifications
You must be signed in to change notification settings - Fork 11
/
inference.py
80 lines (64 loc) · 2.19 KB
/
inference.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
import json
import sys
import os
import shutil
import tensorflow as tf
import numpy as np
from train import build_model
model_name = os.path.join("model", "best_model")
predictions_dir = "predictions"
mispredicted_dir = "mispredicted"
class_names=["cat", "dog", "muffin", "croissant"]
if __name__ == "__main__":
if len(sys.argv) >= 2:
data = sys.argv[1]
else:
print(f"Usage: python {sys.argv[0]} <data directory>")
exit(1)
test = tf.keras.preprocessing.image_dataset_from_directory(
data,
labels=None,
label_mode=None,
image_size=(256, 256),
crop_to_aspect_ratio=True,
shuffle=False
)
model = build_model()
model.load_weights(model_name).expect_partial()
predict = model.predict(test)
pred_indices = np.argmax(predict, -1)
os.mkdir(predictions_dir)
os.mkdir(mispredicted_dir)
for c in class_names:
os.mkdir(os.path.join(mispredicted_dir, c))
count = 0
for index, predicted_class in enumerate(pred_indices):
file_path = test.file_paths[index]
file_name = os.path.basename(file_path)
md5 = os.path.splitext(file_name)[0].split('-')[-1]
file_label = os.path.split(os.path.dirname(file_path))[1]
pred_label = class_names[pred_indices[index]]
confidence = predict[index, pred_indices[index]]
annotation = {
'annotation': {
'inference': {
'label': pred_label,
'confidence': float(confidence)
},
'data-object-info': {
'md5': md5,
'path': file_path
}
}
}
with open(os.path.join(predictions_dir, file_name + '.json'), 'w',) as f:
json.dump(annotation, f, indent=4)
if confidence > 0.97 and file_label != pred_label:
mispredicted_file_name = os.path.join(
mispredicted_dir,
file_label,
f"{pred_label}-{file_name}"
)
shutil.copy(file_path, mispredicted_file_name)
count += 1
print(f"Total mispredicted {count}" )