Skip to content

Commit 4ca346c

Browse files
committed
add label precode code
1 parent 4fe9a8e commit 4ca346c

File tree

1 file changed

+63
-0
lines changed

1 file changed

+63
-0
lines changed

precode.py

Lines changed: 63 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,63 @@
1+
import cv2 as cv
2+
from pathlib import Path
3+
import natsort
4+
import numpy as np
5+
6+
def one_hot_it(label, label_values):
7+
"""
8+
Convert a segmentation image label array to one-hot format
9+
by replacing each pixel value with a vector of length num_classes
10+
11+
# Arguments
12+
label: The 2D array segmentation image label
13+
label_values
14+
15+
# Returns
16+
A 2D array with the same width and hieght as the input, but
17+
with a depth size of num_classes
18+
"""
19+
semantic_map = []
20+
for colour in label_values:
21+
# colour_map = np.full((label.shape[0], label.shape[1], label.shape[2]), colour, dtype=int)
22+
equality = np.equal(label, colour)
23+
class_map = np.all(equality, axis=-1)
24+
semantic_map.append(class_map)
25+
semantic_map = np.stack(semantic_map, axis=-1)
26+
27+
return semantic_map
28+
29+
def reverse_one_hot(image):
30+
"""
31+
Transform a 2D array in one-hot format (depth is num_classes),
32+
to a 2D array with only 1 channel, where each pixel value is
33+
the classified class key.
34+
35+
# Arguments
36+
image: The one-hot format image
37+
38+
# Returns
39+
A 2D array with the same width and hieght as the input, but
40+
with a depth size of 1, where each pixel value is the classified
41+
class key.
42+
"""
43+
x = np.argmax(image, axis=-1)
44+
return x
45+
46+
def load_image(path):
47+
image = cv.cvtColor(cv.imread(path, 1), cv.COLOR_BGR2RGB)
48+
return image
49+
50+
GT_Path = Path("path-to-original-label-images")
51+
GT_File = natsort.natsorted(list(GT_Path.glob("*.png")), alg=natsort.PATH)
52+
GT_Str = []
53+
for i in GT_File:
54+
GT_Str.append(str(i))
55+
56+
out_prefix="precoded_label"
57+
label_values = [[255, 255, 255], [0, 0, 255], [0, 255, 255], [0, 255, 0], [255, 255, 0], [255, 0, 0]]
58+
for k in range(len(GT_Str)):
59+
gt=load_image(GT_Str[k])
60+
out=reverse_one_hot(one_hot_it(gt,label_values))
61+
out_str=out_prefix+Path(GT_Str[k]).name
62+
cv.imwrite(out_str,out)
63+
# print("kk")

0 commit comments

Comments
 (0)