|
a |
|
b/data/data_utils.py |
|
|
1 |
import random |
|
|
2 |
import numpy as np |
|
|
3 |
from PIL import Image |
|
|
4 |
|
|
|
5 |
def read_image(path, dtype=np.float32, color=True): |
|
|
6 |
""" |
|
|
7 |
Read an image from a given path, the image is in (C, H, W) format and the range of its value is between [0, 255] |
|
|
8 |
:param path: The path of an image file |
|
|
9 |
:param dtype: data type of an image, default is float32 |
|
|
10 |
:param color: If 'True', the number of channels is three, in this case, RGB |
|
|
11 |
If 'False', this function returns a gray scale image |
|
|
12 |
:return: |
|
|
13 |
""" |
|
|
14 |
f = Image.open(path) |
|
|
15 |
try: |
|
|
16 |
if color: |
|
|
17 |
img = f.convert('RGB') |
|
|
18 |
else: |
|
|
19 |
img = f.convert('P') |
|
|
20 |
img = np.asarray(img, dtype=dtype) |
|
|
21 |
finally: |
|
|
22 |
if hasattr(f, 'close'): |
|
|
23 |
f.close() |
|
|
24 |
|
|
|
25 |
if img.ndim == 2: |
|
|
26 |
# reshape (H, W) -> (1, H, W) |
|
|
27 |
return img[np.newaxis] |
|
|
28 |
else: |
|
|
29 |
# transpose (H, W, C) -> (C, H, W) |
|
|
30 |
return img.transpose((2, 0, 1)) |
|
|
31 |
|
|
|
32 |
|
|
|
33 |
def flip_masks(masks, y_flip=False, x_flip=False): |
|
|
34 |
masks = masks.copy() |
|
|
35 |
for i in range(masks.shape[0]): |
|
|
36 |
if y_flip: |
|
|
37 |
masks[i] = np.flipud(masks[i]).copy() |
|
|
38 |
if x_flip: |
|
|
39 |
masks[i] = np.fliplr(masks[i]).copy() |
|
|
40 |
return masks |
|
|
41 |
|
|
|
42 |
def random_flip(img, y_random=False, x_random=False, |
|
|
43 |
return_param=False, copy=False): |
|
|
44 |
""" |
|
|
45 |
Randomly flip an image in vertical or horizontal direction. |
|
|
46 |
:param img (numpy.ndarray): An array that gets flipped. |
|
|
47 |
This is in CHW format. |
|
|
48 |
:param y_random (bool): Randomly flip in vertical direction. |
|
|
49 |
:param x_random (bool): Randomly flip in horizontal direction. |
|
|
50 |
:param return_param (bool): Returns information of flip. |
|
|
51 |
:param copy (bool): If False, a view of :obj:`img` will be returned. |
|
|
52 |
:return (numpy.ndarray or (numpy.ndarray, dict)): |
|
|
53 |
If :`return_param = False`, |
|
|
54 |
returns an array :obj:`out_img` that is the result of flipping. |
|
|
55 |
If :obj:`return_param = True`, |
|
|
56 |
returns a tuple whose elements are :obj:`out_img, param`. |
|
|
57 |
:obj:`param` is a dictionary of intermediate parameters whose |
|
|
58 |
contents are listed below with key, value-type and the description |
|
|
59 |
of the value. |
|
|
60 |
* **y_flip** (*bool*): Whether the image was flipped in the\ |
|
|
61 |
vertical direction or not. |
|
|
62 |
* **x_flip** (*bool*): Whether the image was flipped in the\ |
|
|
63 |
horizontal direction or not. |
|
|
64 |
""" |
|
|
65 |
y_flip, x_flip = False, False |
|
|
66 |
if y_random: |
|
|
67 |
y_flip = random.choice([True, False]) |
|
|
68 |
if x_random: |
|
|
69 |
x_flip = random.choice([True, False]) |
|
|
70 |
|
|
|
71 |
if y_flip: |
|
|
72 |
img = img[:, ::-1, :] |
|
|
73 |
if x_flip: |
|
|
74 |
img = img[:, :, ::-1] |
|
|
75 |
|
|
|
76 |
if copy: |
|
|
77 |
img = img.copy() |
|
|
78 |
|
|
|
79 |
if return_param: |
|
|
80 |
return img, {'y_flip': y_flip, 'x_flip': x_flip} |
|
|
81 |
else: |
|
|
82 |
return img |
|
|
83 |
|
|
|
84 |
|
|
|
85 |
if __name__ == '__main__': |
|
|
86 |
pass |