forked from cryu854/FastStyle
-
Notifications
You must be signed in to change notification settings - Fork 0
/
main.py
119 lines (99 loc) · 4.12 KB
/
main.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
# USAGE
# python main.py train --style ./path/to/style/image.jpg(video.mp4) \
# --dataset ./path/to/dataset \
# --weights ./path/to/weights \
# --batch 2
# python main.py evaluate --content ./path/to/content/image.jpg \
# --weights ./path/to/weights \
# --result ./path/to/save/results/image.jpg
import os
import argparse
from train import trainer
from evaluate import transfer
CONTENT_WEIGHT = 6e0
STYLE_WEIGHT = 2e-3
TV_WEIGHT = 6e2
LEARNING_RATE = 1e-3
NUM_EPOCHS = 2
BATCH_SIZE = 2
STYLE_IMAGE = './images/style/udnie.jpg'
CONTENT_IMAGE = './images/content/chicago.jpg'
DATASET_PATH = '../datasets/train2014'
WEIGHTS_PATH = './weights/wave/weights'
RESULT_NAME = './images/results/Result.jpg'
def main():
# Parse command line arguments
parser = argparse.ArgumentParser(
description='Fast Style Transfer')
parser.add_argument('command',
metavar='<command>',
help="'train' or 'evaluate'")
parser.add_argument('--debug', required=False, type=bool,
metavar=False,
help='Whether to print the loss',
default=False)
parser.add_argument('--dataset', required=False,
metavar=DATASET_PATH,
default=DATASET_PATH)
parser.add_argument('--style', required=False,
metavar=STYLE_IMAGE,
help='Style image to train the specific style',
default=STYLE_IMAGE)
parser.add_argument('--content', required=False,
metavar=CONTENT_IMAGE,
help='Content image/video to evaluate with',
default=CONTENT_IMAGE)
parser.add_argument('--weights', required=False,
metavar=WEIGHTS_PATH,
help='Checkpoints directory',
default=WEIGHTS_PATH)
parser.add_argument('--result', required=False,
metavar=RESULT_NAME,
help='Path to the transfer results',
default=RESULT_NAME)
parser.add_argument('--batch', required=False, type=int,
metavar=BATCH_SIZE,
help='Training batch size',
default=BATCH_SIZE)
parser.add_argument('--max_dim', required=False, type=int,
metavar=None,
help='Resize the result image to desired size or remain as the original',
default=None)
args = parser.parse_args()
# Validate arguments
if args.command == "train":
assert os.path.exists(args.dataset), 'dataset path not found !'
assert os.path.exists(args.style), 'style image not found !'
assert args.batch > 0
assert NUM_EPOCHS > 0
assert CONTENT_WEIGHT >= 0
assert STYLE_WEIGHT >= 0
assert TV_WEIGHT >= 0
assert LEARNING_RATE >= 0
parameters = {
'style_file' : args.style,
'dataset_path' : args.dataset,
'weights_path' : args.weights,
'debug' : args.debug,
'content_weight' : CONTENT_WEIGHT,
'style_weight' : STYLE_WEIGHT,
'tv_weight' : TV_WEIGHT,
'learning_rate' : LEARNING_RATE,
'batch_size' : args.batch,
'epochs' : NUM_EPOCHS,
}
trainer(**parameters)
elif args.command == "evaluate":
assert args.content, 'content image/video not found !'
assert args.weights, 'weights path not found !'
parameters = {
'content' : args.content,
'weights' : args.weights,
'max_dim' : args.max_dim,
'result' : args.result,
}
transfer(**parameters)
else:
print('Example usage : python main.py evaluate --content ./path/to/content/image.jpg')
if __name__ == '__main__':
main()