-
Notifications
You must be signed in to change notification settings - Fork 1
/
yolo_aug.py
170 lines (155 loc) · 6.13 KB
/
yolo_aug.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
from random import random
import cv2,time,os,sys
from tqdm import tqdm
from glob import glob
import numpy as np
import pandas as pd
from pathlib import Path
import logging
import getopt
log = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO)
def main(argv):
help_str = 'yolo_aug.py -i <input_dir> -t <aug_type (hflip,vflip,hvflip,bright)> -e <image extension (jpg,jpeg,png ...)> -o <output_dir>'
try:
opts, args = getopt.getopt(
argv,"hi:t:e:o:",["input_dir=","aug_type=", "image extension=" ,"output_dir="])
except getopt.GetoptError:
log.exception(help_str)
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
log.info(help_str)
sys.exit()
elif opt in ("-i", "--input_dir"):
input_dir = arg
elif opt in ("-t", "--aug_type"):
aug_type = arg
elif opt in ("-e", "--image extension"):
image_ext = arg
elif opt in ("-o", "--output_dir"):
output_dir = arg
if aug_type == 'hflip':
output_dir = output_dir + '/hflip_'+ time.strftime("%H%M%S")
out_dir = Path(output_dir)
out_dir.mkdir(parents=True, exist_ok=True)
elif aug_type == 'vflip':
output_dir = output_dir + '/vflip_'+ time.strftime("%H%M%S")
out_dir = Path(output_dir)
out_dir.mkdir(parents=True, exist_ok=True)
elif aug_type == 'hvflip':
output_dir = output_dir + '/hvflip_'+ time.strftime("%H%M%S")
out_dir = Path(output_dir)
out_dir.mkdir(parents=True, exist_ok=True)
elif aug_type == 'bright':
output_dir = output_dir + '/bright_'+ time.strftime("%H%M%S")
out_dir = Path(output_dir)
out_dir.mkdir(parents=True, exist_ok=True)
else:
log.error('Invalid augmentation type')
log.exception(help_str)
sys.exit(2)
log.info('Input directory: {}'.format(input_dir))
log.info('Augmentation type: {}'.format(aug_type))
log.info('Output directory: {}'.format(output_dir))
# --------- read yolo format label file -------------
def boxesFromYOLO(imagePath,labelPath):
image = cv2.imread(imagePath)
(hI, wI) = image.shape[:2]
lines = [line.rstrip('\n') for line in open(labelPath)]
boxes = []
if lines != ['']:
for line in lines:
components = line.split(" ")
category = components[0]
x = int(float(components[1])*wI - float(components[3])*wI/2)
y = int(float(components[2])*hI - float(components[4])*hI/2)
h = int(float(components[4])*hI)
w = int(float(components[3])*wI)
boxes.append((category, (x, y, w, h)))
return (hI, wI,image,boxes)
# -------- vertical flip function --------
def flip_ver(image,boxes,H,W):
txt_yolo = []
img= cv2.flip(image,1)
(H,W) = img.shape[:2]
for box in boxes:
class_name = int(box[0])
(x, y, w, h) = box[1]
x2 = ((W - x -w)+w/2)/W
h2 = h/H
w2 = w/W
y2 = (y+(h/2))/H
txt_yolo.append((class_name,round(x2,4),round(y2,4),round(w2,4),round(h2,4)))
return img,txt_yolo
# -------- horizontal and vertical flip function --------
def flip_hor_ver(image,boxes,H,W):
txt_yolo = []
img= cv2.flip(image,-1)
(H,W) = img.shape[:2]
for box in boxes:
class_name =int(box[0])
(x, y, w, h) = box[1]
x2 = ((W - x -w)+w/2)/W
h2 = h/H
w2 = w/W
y2 =((H -y -h)+h/2)/H
txt_yolo.append((class_name,round(x2,4),round(y2,4),round(w2,4),round(h2,4)))
return img,txt_yolo
# -------- Horizontal flip function --------
def flip_hor(image,boxes,H,W):
txt_yolo = []
img= cv2.flip(image,0)
(H,W) = img.shape[:2]
for box in boxes:
class_name =int(box[0])
(x, y, w, h) = box[1]
x2 = (x+(w/2))/W
h2 = h/H
w2 = w/W
y2 =((H -y -h)+h/2)/H
txt_yolo.append((class_name,round(x2,4),round(y2,4),round(w2,4),round(h2,4)))
return img,txt_yolo
# -------- Brightness function --------
def brightness_augment(img,boxes,H,W):
hsv = cv2.cvtColor(img, cv2.COLOR_RGB2HSV) #convert to hsv
hsv = np.array(hsv, dtype=np.float64)
factor = random()
hsv[:, :, 2] = hsv[:, :, 2] * (factor + np.random.uniform()) #scale channel V uniformly
hsv[:, :, 2][hsv[:, :, 2] > 255] = 255 #reset out of range values
rgb = cv2.cvtColor(np.array(hsv, dtype=np.uint8), cv2.COLOR_HSV2RGB)
(H,W) = img.shape[:2]
txt_yolo = []
for box in boxes:
class_name =int(box[0])
(x, y, w, h) = box[1]
x2 = (x+(w/2))/W
h2 = h/H
w2 = w/W
y2 =(y+(h/2))/H
txt_yolo.append((class_name,round(x2,4),round(y2,4),round(w2,4),round(h2,4)))
return rgb,txt_yolo
# -------- read image and label file -------------
images = glob(input_dir + f'/*.{image_ext}')
for image_path in tqdm(images):
(hI, wI,image,boxes) = boxesFromYOLO(image_path,image_path.replace(image_ext, 'txt'))
if aug_type == 'hflip':
img,txt_yolo = flip_hor(image,boxes,hI,wI)
elif aug_type == 'vflip':
img,txt_yolo = flip_ver(image,boxes,hI,wI)
elif aug_type == 'hvflip':
img,txt_yolo = flip_hor_ver(image,boxes,hI,wI)
elif aug_type == 'bright':
img,txt_yolo = brightness_augment(image,boxes,hI,wI)
else:
log.error('Unknown augmentation type: {}'.format(aug_type))
sys.exit(2)
cv2.imwrite(output_dir + '/' + image_path.split('/')[-1], img)
with open(output_dir + '/' + image_path.split('/')[-1].replace(image_ext, 'txt'), 'w') as f:
if txt_yolo != []:
for box in txt_yolo:
f.write(f'{box[0]} {box[1]} {box[2]} {box[3]} {box[4]}\n')
else:
f.write('')
if __name__ == '__main__':
main(sys.argv[1:])