Spaces:
Running
Running
Update utils.py
Browse files
utils.py
CHANGED
|
@@ -1,40 +1,42 @@
|
|
| 1 |
-
from paddleocr import PaddleOCR
|
| 2 |
-
from PIL import Image
|
| 3 |
-
from numpy import
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
int(1000 * (bbox[
|
| 11 |
-
int(1000 * (bbox[
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
width * (bbox[
|
| 20 |
-
height * (bbox[
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
# print(line[
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
|
|
|
|
|
|
|
|
| 1 |
+
from paddleocr import PaddleOCR
|
| 2 |
+
from PIL import Image
|
| 3 |
+
from numpy import asarrayù
|
| 4 |
+
|
| 5 |
+
ocr = PaddleOCR(use_angle_cls=True)
|
| 6 |
+
|
| 7 |
+
def normalize_bbox(bbox, width, height):
|
| 8 |
+
|
| 9 |
+
return [
|
| 10 |
+
int(1000 * (bbox[0] / width)),
|
| 11 |
+
int(1000 * (bbox[1] / height)),
|
| 12 |
+
int(1000 * (bbox[2] / width)),
|
| 13 |
+
int(1000 * (bbox[3] / height)),
|
| 14 |
+
]
|
| 15 |
+
|
| 16 |
+
def unnormalize_box(bbox, width, height):
|
| 17 |
+
|
| 18 |
+
return [
|
| 19 |
+
width * (bbox[0] / 1000),
|
| 20 |
+
height * (bbox[1] / 1000),
|
| 21 |
+
width * (bbox[2] / 1000),
|
| 22 |
+
height * (bbox[3] / 1000),
|
| 23 |
+
]
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
def OCR(image):
|
| 27 |
+
|
| 28 |
+
result = ocr.ocr(asarray(image), cls=True)
|
| 29 |
+
bboxes = []
|
| 30 |
+
words = []
|
| 31 |
+
|
| 32 |
+
for idx in range(len(result)):
|
| 33 |
+
res = result[idx]
|
| 34 |
+
|
| 35 |
+
for line in res:
|
| 36 |
+
# print(line)
|
| 37 |
+
# print(line[0][0] + line[0][2])
|
| 38 |
+
bboxes.append(normalize_bbox(line[0][0]+line[0][2], image.width, image.height))
|
| 39 |
+
# print(line[1][0])
|
| 40 |
+
words.append(line[1][0])
|
| 41 |
+
|
| 42 |
+
return bboxes, words
|