Update app.py
Browse files
app.py
CHANGED
@@ -23,13 +23,19 @@ Paddle OCR
|
|
23 |
"""
|
24 |
def ocr_with_paddle(img):
|
25 |
finaltext = ''
|
26 |
-
ocr = PaddleOCR(lang='vi',
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
return finaltext
|
34 |
|
35 |
"""
|
@@ -37,13 +43,12 @@ Keras OCR
|
|
37 |
"""
|
38 |
def ocr_with_keras(img):
|
39 |
output_text = ''
|
40 |
-
pipeline=keras_ocr.pipeline.Pipeline()
|
41 |
-
images=[keras_ocr.tools.read(img)]
|
42 |
-
predictions=pipeline.recognize(images)
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
return output_text
|
47 |
|
48 |
"""
|
49 |
easy OCR
|
@@ -56,38 +61,37 @@ def get_grayscale(image):
|
|
56 |
def thresholding(src):
|
57 |
return cv2.threshold(src,127,255, cv2.THRESH_TOZERO)[1]
|
58 |
def ocr_with_easy(img):
|
59 |
-
gray_scale_image=get_grayscale(img)
|
60 |
thresholding(gray_scale_image)
|
61 |
-
cv2.imwrite('image.png',gray_scale_image)
|
62 |
-
reader = easyocr.Reader(['vi'])
|
63 |
-
bounds = reader.readtext('image.png',paragraph=
|
64 |
-
|
65 |
-
return
|
66 |
|
67 |
"""
|
68 |
Generate OCR
|
69 |
"""
|
70 |
-
def generate_ocr(Method,img):
|
71 |
-
|
|
|
|
|
72 |
text_output = ''
|
73 |
-
|
74 |
-
|
75 |
-
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
return text_output
|
89 |
-
else:
|
90 |
-
raise gr.Error("Please upload an image!!!!")
|
91 |
|
92 |
# except Exception as e:
|
93 |
# print("Error in ocr generation ==>",e)
|
|
|
23 |
"""
|
24 |
def ocr_with_paddle(img):
|
25 |
finaltext = ''
|
26 |
+
ocr = PaddleOCR(lang='en') # Không hỗ trợ đa ngôn ngữ kiểu 'vi|en', phải chạy 2 lần
|
27 |
+
result_en = ocr.ocr(img)
|
28 |
+
|
29 |
+
ocr_vi = PaddleOCR(lang='vi')
|
30 |
+
result_vi = ocr_vi.ocr(img)
|
31 |
+
|
32 |
+
def extract_text(result):
|
33 |
+
return ' '.join([line[1][0] for line in result[0]])
|
34 |
+
|
35 |
+
en_text = extract_text(result_en)
|
36 |
+
vi_text = extract_text(result_vi)
|
37 |
+
|
38 |
+
finaltext = f"[EN]: {en_text}\n[VI]: {vi_text}"
|
39 |
return finaltext
|
40 |
|
41 |
"""
|
|
|
43 |
"""
|
44 |
def ocr_with_keras(img):
|
45 |
output_text = ''
|
46 |
+
pipeline = keras_ocr.pipeline.Pipeline()
|
47 |
+
images = [keras_ocr.tools.read(img)]
|
48 |
+
predictions = pipeline.recognize(images)
|
49 |
+
for text, box in predictions[0]:
|
50 |
+
output_text += ' ' + text
|
51 |
+
return "[Detected]: " + output_text
|
|
|
52 |
|
53 |
"""
|
54 |
easy OCR
|
|
|
61 |
def thresholding(src):
|
62 |
return cv2.threshold(src,127,255, cv2.THRESH_TOZERO)[1]
|
63 |
def ocr_with_easy(img):
|
64 |
+
gray_scale_image = get_grayscale(img)
|
65 |
thresholding(gray_scale_image)
|
66 |
+
cv2.imwrite('image.png', gray_scale_image)
|
67 |
+
reader = easyocr.Reader(['vi', 'en']) # Hỗ trợ tiếng Việt và tiếng Anh
|
68 |
+
bounds = reader.readtext('image.png', paragraph=False, detail=0)
|
69 |
+
result_text = '\n'.join(bounds)
|
70 |
+
return result_text
|
71 |
|
72 |
"""
|
73 |
Generate OCR
|
74 |
"""
|
75 |
+
def generate_ocr(Method, img):
|
76 |
+
if img is None or not (img).any():
|
77 |
+
raise gr.Error("Please upload an image!")
|
78 |
+
|
79 |
text_output = ''
|
80 |
+
print("Method selected:", Method)
|
81 |
+
|
82 |
+
if Method == 'EasyOCR':
|
83 |
+
text_output = ocr_with_easy(img)
|
84 |
+
elif Method == 'KerasOCR':
|
85 |
+
text_output = ocr_with_keras(img)
|
86 |
+
elif Method == 'PaddleOCR':
|
87 |
+
text_output = ocr_with_paddle(img)
|
88 |
+
|
89 |
+
try:
|
90 |
+
flag(Method, text_output, img)
|
91 |
+
except Exception as e:
|
92 |
+
print("Flag error:", e)
|
93 |
+
|
94 |
+
return text_output
|
|
|
|
|
|
|
95 |
|
96 |
# except Exception as e:
|
97 |
# print("Error in ocr generation ==>",e)
|