Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -72,18 +72,23 @@ def ocr_lang(lang_list):
|
|
| 72 |
return lang_str
|
| 73 |
|
| 74 |
|
| 75 |
-
import pytesseract
|
| 76 |
-
|
| 77 |
|
| 78 |
# Set Tesseract executable path in Colab virtal environment
|
| 79 |
-
pytesseract.pytesseract.tesseract_cmd = "/usr/bin/tesseract"
|
| 80 |
|
| 81 |
# Set up the Tesseract data directory
|
| 82 |
-
os.environ["TESSDATA_PREFIX"] = "/usr/share/tesseract-ocr/4.00/tessdata"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
|
|
|
|
| 84 |
def ocr_tesseract(img, languages):
|
| 85 |
-
|
| 86 |
-
ocr_str = pytesseract.image_to_string(img, config=custom_config)
|
| 87 |
return ocr_str
|
| 88 |
|
| 89 |
|
|
|
|
| 72 |
return lang_str
|
| 73 |
|
| 74 |
|
| 75 |
+
#import pytesseract
|
| 76 |
+
|
| 77 |
|
| 78 |
# Set Tesseract executable path in Colab virtal environment
|
| 79 |
+
#pytesseract.pytesseract.tesseract_cmd = "/usr/bin/tesseract"
|
| 80 |
|
| 81 |
# Set up the Tesseract data directory
|
| 82 |
+
#os.environ["TESSDATA_PREFIX"] = "/usr/share/tesseract-ocr/4.00/tessdata"
|
| 83 |
+
|
| 84 |
+
#def ocr_tesseract(img, languages):
|
| 85 |
+
# custom_config = f'--oem 3 --psm 6 -l {ocr_lang(languages)}'
|
| 86 |
+
# ocr_str = pytesseract.image_to_string(img, config=custom_config)
|
| 87 |
+
# return ocr_str
|
| 88 |
|
| 89 |
+
# ocr tesseract
|
| 90 |
def ocr_tesseract(img, languages):
|
| 91 |
+
ocr_str = pytesseract.image_to_string(img, lang=ocr_lang(languages))
|
|
|
|
| 92 |
return ocr_str
|
| 93 |
|
| 94 |
|