Spaces:
Sleeping
Sleeping
Delete utils.py
Browse files
utils.py
DELETED
@@ -1,40 +0,0 @@
|
|
1 |
-
from paddleocr import PaddleOCR
|
2 |
-
from PIL import Image
|
3 |
-
|
4 |
-
def normalize_bbox(bbox, width, height):
|
5 |
-
|
6 |
-
return [
|
7 |
-
int(1000 * (bbox[0] / width)),
|
8 |
-
int(1000 * (bbox[1] / height)),
|
9 |
-
int(1000 * (bbox[2] / width)),
|
10 |
-
int(1000 * (bbox[3] / height)),
|
11 |
-
]
|
12 |
-
|
13 |
-
def unnormalize_box(bbox, width, height):
|
14 |
-
|
15 |
-
return [
|
16 |
-
width * (bbox[0] / 1000),
|
17 |
-
height * (bbox[1] / 1000),
|
18 |
-
width * (bbox[2] / 1000),
|
19 |
-
height * (bbox[3] / 1000),
|
20 |
-
]
|
21 |
-
|
22 |
-
|
23 |
-
def OCR(image_path: str):
|
24 |
-
ocr = PaddleOCR(use_angle_cls=True)
|
25 |
-
image = Image.open(image_path)
|
26 |
-
result = ocr.ocr(image_path, cls=True)
|
27 |
-
bboxes = []
|
28 |
-
words = []
|
29 |
-
|
30 |
-
for idx in range(len(result)):
|
31 |
-
res = result[idx]
|
32 |
-
|
33 |
-
for line in res:
|
34 |
-
# print(line)
|
35 |
-
# print(line[0][0] + line[0][2])
|
36 |
-
bboxes.append(normalize_bbox(line[0][0]+line[0][2], image.width, image.height))
|
37 |
-
# print(line[1][0])
|
38 |
-
words.append(line[1][0])
|
39 |
-
|
40 |
-
return bboxes, words
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|