python使用Augmentor对图像分割数据集中原图和标签mask同时进行变换

$$python使用Augmentor对图像分割数据集中原图和标签mask同时进行变换$$

import Augmentor

'''
步骤 1.创建管道Pipeline实例
'''
# 设置原图路径
p = Augmentor.Pipeline("defects")
# 设置标签mask路径
p.ground_truth("rgb_masks")

'''
步骤 2：向管道添加数据增强（执行概率+参数范围）操作
'''
# 设置处理概率
process_probability = 1

# 1、旋转
# 1.1 不带裁剪的旋转
p.rotate_without_crop(probability=process_probability, max_left_rotation=60, max_right_rotation=60)
# 1.2 带裁剪的旋转
p.rotate(probability=process_probability, max_left_rotation=25, max_right_rotation=25)
# 1.3 旋转90°
p.rotate_random_90(probability=process_probability)

# 2、缩放
p.zoom(probability=process_probability, min_factor=0.5, max_factor=2)

# 3、裁剪
# 3.1 按大小裁剪
# p.crop_by_size(probability=1, width=1000, height=1000)
# 3.2 从区域中心裁剪
p.crop_centre(probability=1,percentage_area=0.8)

# 3.3 随机裁剪
p.crop_random(probability=1, percentage_area=0.8, randomise_percentage_area=True)

# 4、翻转
# 4.1 水平翻转
p.flip_left_right(probability=process_probability)

# 4.2 上下翻转
p.flip_top_bottom(probability=process_probability)

# 4.3 随机翻转
p.flip_random(probability=process_probability)

# 5、亮度增强/减弱
p.random_brightness(probability=process_probability, min_factor=0.7, max_factor=1.2)

# 6、颜色增强/减弱
p.random_color(probability=process_probability, min_factor=0.0, max_factor=1.5)

# 7、对比度增强/减弱
p.random_contrast(probability=process_probability, min_factor=0.7, max_factor=1.2)

# 8、错切形变
p.shear(probability=process_probability,max_shear_left=15,max_shear_right=15)

# 9、透视形变
# 9.1 垂直方向透视形变
p.skew_tilt(probability=process_probability, magnitude=1)

# 9.2 斜四角方向透视形变
p.skew_corner(probability=process_probability, magnitude=0.5)

# 10、弹性扭曲
p.random_distortion(probability=process_probability, grid_height=5, grid_width=16, magnitude=8)

# 11、随机区域擦除
p.random_erasing(probability=process_probability, rectangle_area=0.5)  


'''
步骤 3：生成数据增强后的图像和标签mask
'''
# 设置生成个数
p.sample(100)
# 对每个图像做一次数据增强操作
p.process()
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

一 使用步骤

步骤 1.创建管道Pipeline实例

import Augmentor
# 设置原图路径
p = Augmentor.Pipeline("./defects")
# 设置标签mask路径
p.ground_truth("./rgb_masks")
1
2
3
4
5

步骤 2：向管道添加数据增强（执行概率+参数范围）操作

'''
步骤 2：向管道添加数据增强（执行概率+参数范围）操作
'''
# 设置处理概率
process_probability = 1

# 1、旋转
# 1.1 不带裁剪的旋转
p.rotate_without_crop(probability=process_probability, max_left_rotation=60, max_right_rotation=60)
# 1.2 带裁剪的旋转
p.rotate(probability=process_probability, max_left_rotation=25, max_right_rotation=25)
# 1.3 旋转90°
p.rotate_random_90(probability=process_probability)

# 2、缩放
p.zoom(probability=process_probability, min_factor=0.5, max_factor=2)

# 3、裁剪
# 3.1 按大小裁剪
# p.crop_by_size(probability=1, width=1000, height=1000)
# 3.2 从区域中心裁剪
p.crop_centre(probability=1,percentage_area=0.8)

# 3.3 随机裁剪
p.crop_random(probability=1, percentage_area=0.8, randomise_percentage_area=True)

# 4、翻转
# 4.1 水平翻转
p.flip_left_right(probability=process_probability)

# 4.2 上下翻转
p.flip_top_bottom(probability=process_probability)

# 4.3 随机翻转
p.flip_random(probability=process_probability)

# 5、亮度增强/减弱
p.random_brightness(probability=process_probability, min_factor=0.7, max_factor=1.2)

# 6、颜色增强/减弱
p.random_color(probability=process_probability, min_factor=0.0, max_factor=1.5)

# 7、对比度增强/减弱
p.random_contrast(probability=process_probability, min_factor=0.7, max_factor=1.2)

# 8、错切形变
p.shear(probability=process_probability,max_shear_left=15,max_shear_right=15)

# 9、透视形变
# 9.1 垂直方向透视形变
p.skew_tilt(probability=process_probability, magnitude=1)

# 9.2 斜四角方向透视形变
p.skew_corner(probability=process_probability, magnitude=0.5)

# 10、弹性扭曲
p.random_distortion(probability=process_probability, grid_height=5, grid_width=16, magnitude=8)

# 11、随机区域擦除
p.random_erasing(probability=process_probability, rectangle_area=0.5)

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

步骤 3：生成数据增强后的图像和标签mask

# 设置生成个数
p.sample(5)
# 对每个图像做一次数据增强操作
p.process()  
1
2
3
4

新生成的增强图像和标签mask会保存到图像数据集的目录下名为 output 的目录中

---

二 Augmentor数据增强详解

1、旋转

2、缩放

3、裁剪

4、翻转

5、亮度增强/减弱

6、颜色增强/减弱

7、对比度增强/减弱

8、错切形变

9、透视形变

10、弹性扭曲

11、随机区域擦除

示例

import Augmentor

'''
步骤 1.创建管道Pipeline实例
'''
# 设置原图路径
p = Augmentor.Pipeline("defects")
# 设置标签mask路径
p.ground_truth("rgb_masks")
# 设置生成增强数据样本数量
sample_num = 10
'''
设置标志位
'''
# 1、旋转
is_rotate = True

# 2、缩放
is_zoom = True

# 3、裁剪
is_crop = False

# 4、翻转
is_flip = False

# 5、亮度增强/减弱
is_brightness = True

# 6、颜色增强/减弱
is_color = True

# 7、对比度增强/减弱
is_contrast = False

# 8、错切形变
is_sheer = False

# 9、透视形变
is_skew = False

# 10、弹性扭曲
is_distortion = False

# 11、随机区域擦除
is_erasing = False

'''
设置参数范围
'''

# 1、旋转
rotate_max_left_rotation = 60
rotate_max_right_rotation = 60
# 2、缩放
zoom_min_factor = 0.5
zoom_max_factor = 2
# 3、裁剪
crop_percentage_area=0.8
# 4、翻转

# 5、亮度增强/减弱
brightness_min_factor=0.7  
brightness_max_factor=1.2
# 6、颜色增强/减弱
color_min_factor=0.0    
color_max_factor=1.5
# 7、对比度增强/减弱
contrast_min_factor=0.7  
contrast_max_factor=1.2
# 8、错切形变
sheer_max_shear_left=15  
sheer_max_shear_right=15
# 9、透视形变
skew_tilt_magnitude = 1  
skew_corner_magnitude = 0.5
# 10、弹性扭曲
distortion_grid_height=5 
distortion_grid_width=16  
distortion_magnitude=8
# 11、随机区域擦除
erasing_rectangle_area = 0.5


'''
步骤 2：向管道添加数据增强（执行概率+参数范围）操作
'''
# 设置处理概率
process_probability = 1
# 1、旋转
if is_rotate:
    # 1.1 不带裁剪的旋转
    p.rotate_without_crop(probability=process_probability, max_left_rotation=rotate_max_left_rotation, max_right_rotation=rotate_max_right_rotation)
    # # 1.2 带裁剪的旋转
    # p.rotate(probability=process_probability, max_left_rotation=25, max_right_rotation=25)
    # # 1.3 旋转90°
    # p.rotate_random_90(probability=process_probability)

# 2、缩放   zoom_min_factor=0.5   zoom_max_factor=2
if is_zoom:
    p.zoom(probability=process_probability, min_factor=zoom_min_factor, max_factor=zoom_max_factor)

# 3、裁剪   crop_percentage_area=0.8
if is_crop:
    # 3.1 按大小裁剪
    # p.crop_by_size(probability=1, width=1000, height=1000)
    # 3.2 从区域中心裁剪
    # p.crop_centre(probability=1,percentage_area=0.8)
    # 3.3 随机裁剪
    p.crop_random(probability=process_probability, percentage_area=crop_percentage_area, randomise_percentage_area=True)

# 4、翻转
if is_flip:
    # # 4.1 水平翻转
    # p.flip_left_right(probability=process_probability)
    #
    # # 4.2 上下翻转
    # p.flip_top_bottom(probability=process_probability)
    # 4.3 随机翻转
    p.flip_random(probability=process_probability)

# 5、亮度增强/减弱   brightness_min_factor=0.7  brightness_max_factor=1.2
if is_brightness:
    p.random_brightness(probability=process_probability, min_factor=brightness_min_factor, max_factor=brightness_max_factor)

# 6、颜色增强/减弱  color_min_factor=0.0    color_max_factor=1.5
if is_color:
    p.random_color(probability=process_probability, min_factor=color_min_factor, max_factor=color_max_factor)

# 7、对比度增强/减弱  contrast_min_factor=0.7  contrast_max_factor=1.2
if is_contrast:
    p.random_contrast(probability=process_probability, min_factor=contrast_min_factor, max_factor=contrast_max_factor)

# 8、错切形变  sheer_max_shear_left=15  sheer_max_shear_right=15
if is_sheer:
    p.shear(probability=process_probability,max_shear_left=sheer_max_shear_left,max_shear_right=sheer_max_shear_right)

# 9、透视形变  skew_tilt_magnitude = 1  skew_corner_magnitude = 0.5
if is_skew:
    # 9.1 垂直方向透视形变
    p.skew_tilt(probability=process_probability, magnitude=skew_tilt_magnitude)
    # 9.2 斜四角方向透视形变
    p.skew_corner(probability=process_probability, magnitude=skew_corner_magnitude)

# 10、弹性扭曲   distortion_grid_height=5 distortion_grid_width=16  distortion_magnitude=8
if is_distortion:
    p.random_distortion(probability=process_probability, grid_height=distortion_grid_height, grid_width=distortion_grid_width, magnitude=distortion_magnitude)

# 11、随机区域擦除  erasing_rectangle_area = 0.5
if is_erasing:
    p.random_erasing(probability=process_probability, rectangle_area=erasing_rectangle_area)


'''
步骤 3：生成数据增强后的图像和标签mask
'''
# 设置生成个数
p.sample(sample_num)
# 对每个图像做一次数据增强操作
p.process()
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