image_to_data recognize 'conf' as string

[Brunitus Nishimura]

Could anyone explain why tesseract recognizes "conf" as a string ?

1. import pytesseract
2. import cv2
3. from pytesseract import Output
4.  
5. pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"
6. img = cv2.imread(r"C:\Users\Documents\Python\OCR\Programa\teste_manuscrito_01.jpg")
7. rgb = cv2.cvtColor (img, cv2.COLOR_BGR2RGB)
8.  
9. resultado = pytesseract.image_to_data(rgb, lang='por', output_type=Output.DICT)
10.  
11. print(resultado)
12.  
13. {'level': [1, 2, 3, 4, 5, 4, 5, 5],
14. 'page_num': [1, 1, 1, 1, 1, 1, 1, 1],
15. 'block_num': [0, 1, 1, 1, 1, 1, 1, 1],
16. 'par_num': [0, 0, 1, 1, 1, 1, 1, 1],
17. 'line_num': [0, 0, 0, 1, 1, 2, 2, 2],
18. 'word_num': [0, 0, 0, 0, 1, 0, 1, 2],
19. 'left': [0, 38, 38, 38, 38, 102, 102, 307],
20. 'top': [0, 79, 79, 79, 79, 228, 233, 228],
21. 'width': [700, 607, 607, 607, 607, 532, 77, 327],
22. 'height': [400, 236, 236, 92, 92, 87, 76, 87],
23. 'conf': ['-1', '-1', '-1', '-1', '90.214363', '-1', '77.749153', '61.677670'],
24. 'text': ['', '', '', '', 'TESTANDO', '', 'O', 'OCR...']}

///////////////////////////////////////////

confianca = int(resultado['conf'] [i])

ValueError: invalid literal for int() with base 10: '90.214363'

[Zdenko Podobny]

It is not the tesseract that produces conf as string but pytesseract.

You can not convert float string to int directly in python. Check your python tutorial how to do it correctly.

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/6339b18a-95a6-4085-abff-7a0e25af4cd4n%40googlegroups.com.

[Quang Linh]

you can change int() to float() 

Vào lúc 14:32:58 UTC+7 ngày Thứ Sáu, 17 tháng 9, 2021, zdenop đã viết:

[Mohammad Waqas Shoukat Ali]

Can someone guide me how to enable Regix inside tesseract output to capture certain keywords ? What is best approach to catch certain keywords and its corresponding values, Any advise or help really appreciated. 

Regards,

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/0462e4a5-a919-4039-bf53-2c6f0d006d51n%40googlegroups.com.