-
Notifications
You must be signed in to change notification settings - Fork 1
/
strong_transform.py
37 lines (36 loc) · 1.34 KB
/
strong_transform.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
from torchvision import transforms
from albumentations import (
HorizontalFlip, IAAPerspective, ShiftScaleRotate, CLAHE, RandomRotate90,JpegCompression,
Transpose, ShiftScaleRotate, Blur,GaussianBlur, OpticalDistortion, GridDistortion, HueSaturationValue,
IAAAdditiveGaussianNoise, GaussNoise, MotionBlur, MedianBlur, IAAPiecewiseAffine,
IAASharpen, IAAEmboss, RandomBrightnessContrast, Flip, OneOf, Compose,RandomBrightness,ToSepia
)
import numpy as np
def strong_aug(p=0.5):
return Compose([
HorizontalFlip(),
OneOf([
IAAAdditiveGaussianNoise(),
GaussNoise(),
], p=0.2),
OneOf([
MotionBlur(p=0.25),
JpegCompression(50,90),
GaussianBlur(p=0.5),
Blur(blur_limit=3, p=0.25),
], p=0.2),
HueSaturationValue(p=0.2),
OneOf([
RandomBrightness(),
IAASharpen(),
IAAEmboss(),
RandomBrightnessContrast(),
], p=0.6),
ToSepia(p=0.1)
], p=p)
augmentation=strong_aug()
trans=transforms.Compose([
transforms.ToTensor(),
# transforms.Normalize(mean=[0.38348666],std=[0.20834281])
transforms.Normalize(mean=[0.38348666, 0.39193852, 0.4665315],std=[0.20834281, 0.20540032, 0.24183848])
])