Spaces:
Runtime error
Runtime error
Commit ·
82700d7
1
Parent(s): c6826bf
add tesseract-ocr
Browse files- app/__pycache__/ocr.cpython-310.pyc +0 -0
- app/ocr.py +9 -2
- temp_KTP Evan.pdf +0 -0
- temp_WhatsApp Image 2023-08-04 at 10.48.42.jpeg +0 -0
- temp_page_0.png +0 -0
app/__pycache__/ocr.cpython-310.pyc
CHANGED
|
Binary files a/app/__pycache__/ocr.cpython-310.pyc and b/app/__pycache__/ocr.cpython-310.pyc differ
|
|
|
app/ocr.py
CHANGED
|
@@ -3,7 +3,7 @@ import fitz
|
|
| 3 |
import pytesseract
|
| 4 |
from PIL import Image
|
| 5 |
import cv2
|
| 6 |
-
import io
|
| 7 |
import pymupdf
|
| 8 |
from fastapi import HTTPException
|
| 9 |
import numpy as np
|
|
@@ -11,7 +11,14 @@ from pytesseract import Output
|
|
| 11 |
import imutils
|
| 12 |
|
| 13 |
|
| 14 |
-
pytesseract.pytesseract.tesseract_cmd = r'C:\\Program Files\\Tesseract-OCR\\tesseract.exe'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
def reformat_ocr_result(result):
|
| 17 |
mapping = {
|
|
|
|
| 3 |
import pytesseract
|
| 4 |
from PIL import Image
|
| 5 |
import cv2
|
| 6 |
+
import io, os
|
| 7 |
import pymupdf
|
| 8 |
from fastapi import HTTPException
|
| 9 |
import numpy as np
|
|
|
|
| 11 |
import imutils
|
| 12 |
|
| 13 |
|
| 14 |
+
# pytesseract.pytesseract.tesseract_cmd = r'C:\\Program Files\\Tesseract-OCR\\tesseract.exe'
|
| 15 |
+
pytesseract.pytesseract.tesseract_cmd = 'tesseract'
|
| 16 |
+
# if os.name == 'nt':
|
| 17 |
+
# # Windows
|
| 18 |
+
# pytesseract.pytesseract.tesseract_cmd = r'C:\\Program Files\\Tesseract-OCR\\tesseract.exe'
|
| 19 |
+
# else:
|
| 20 |
+
# # Unix/Linux, where tesseract should be in the PATH
|
| 21 |
+
# pytesseract.pytesseract.tesseract_cmd = 'tesseract'
|
| 22 |
|
| 23 |
def reformat_ocr_result(result):
|
| 24 |
mapping = {
|
temp_KTP Evan.pdf
ADDED
|
Binary file (932 kB). View file
|
|
|
temp_WhatsApp Image 2023-08-04 at 10.48.42.jpeg
ADDED
|
temp_page_0.png
ADDED
|