-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_preprocess.py
70 lines (59 loc) · 2.57 KB
/
data_preprocess.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
#Import Libraries
print("Importing libraries...")
import cv2 as cv
import os, os.path
import numpy as np
import random
import imutils
print("Libraries imported")
#Directories
HOME_DIR = os.getcwd()
PICS_DIR = os.path.join(HOME_DIR, 'Pictures') #Unprocessed data
TRAINDATA_DIR = os.path.join(HOME_DIR, 'Data/Train')
#Image Augmentation
def dataAugment(img, label, count):
img = cv.resize(img, (224,224))
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '.png'), img)
img_gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY) #Convert RGB image to Grayscale
img_gray_3ch = np.zeros_like(img)
img_gray_3ch[:,:,0] = img_gray
img_gray_3ch[:,:,1] = img_gray
img_gray_3ch[:,:,2] = img_gray
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '_gray.png'), img_gray_3ch)
#Add speckle noise
row,col,ch = img.shape
gauss = np.random.randn(row,col,ch)
gauss = gauss.reshape(row,col,ch)
img_noise = img + img * gauss
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '_noise.png'), img_noise)
img_flip = cv.flip(img, 1) #Flip horizontally
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '_flip.png'), img_flip)
angle = random.choice([-90,-75,-60,-45,-30,-15,15,30,45,60,75,90]) #Choose random angle from given angles
img_rotate = imutils.rotate(img, angle)
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '_rot.png'), img_rotate)
if __name__ == "__main__":
print("Processing images...")
dataAugment_flag = input("Do you want to perform data augmentation?: Enter y/n \n")
#Preprocess all the images of Class 0
count = 0
label = 'neg'
for img_name in os.listdir(os.path.join(PICS_DIR, label)):
count += 1
img = cv.imread(os.path.join(PICS_DIR, label, img_name))
if(dataAugment_flag == 'y' or dataAugment_flag == 'Y'):
dataAugment(img, label, count)
else:
img = cv.resize(img, (224,224))
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '.png'), img)
#Preprocess all the images of Class 1
count = 0
label = 'pos'
for img_name in os.listdir(os.path.join(PICS_DIR, label)):
count += 1
img = cv.imread(os.path.join(PICS_DIR, label, img_name))
if(dataAugment_flag == 'y' or dataAugment_flag == 'Y'):
dataAugment(img, label, count)
else:
img = cv.resize(img, (224,224))
cv.imwrite(os.path.join(TRAINDATA_DIR, label +'_'+ str(count) + '.png'), img)
print("Processing completed")