worked pdf images -> text worked

 

orignial

"""!pip install pdf2image

from PIL import Image

import pytesseract

import os

from pdf2image import convert_from_path

# Paths to the files

pdf_file_path = '/content/Pravakthalu-Yevaru.pdf'

output_text_file_path = '/content/Pravakthalu-Yevaru.txt'

# Verify the PDF file existence

if not os.path.exists(pdf_file_path):

    raise FileNotFoundError(f"PDF file {pdf_file_path} not found")

# Convert PDF to images

images = convert_from_path(pdf_file_path, dpi=300)

# Configure Tesseract for Telugu and English

tesseract_config = r'--oem 3 --psm 6 -l tel+eng'

# Process each image and save the text

with open(output_text_file_path, 'w', encoding='utf-8') as output_file:

    for i, image in enumerate(images):

        # Extract text from the image using Tesseract

        text = pytesseract.image_to_string(image, config=tesseract_config)

        output_file.write(f"Page {i + 1}\n")

        output_file.write(text)

        output_file.write("\n\n")

print(f'OCR text written to file "{output_text_file_path}"')"""

#it worked, see below optmized code,

!pip install Pillow

!pip install pytesseract

!apt-get install -y poppler-utils

!pip install pdf2image

!apt-get update

!apt-get install -y tesseract-ocr-tel

from PIL import Image

import pytesseract

import os

from pdf2image import convert_from_path

# Paths to the files

pdf_file_path = '/content/Images_filename123.pdf'

output_text_file_path = '/content/out.txt'

# Verify the PDF file existence

if not os.path.exists(pdf_file_path):

    raise FileNotFoundError(f"PDF file {pdf_file_path} not found")

# Convert PDF to images

images = convert_from_path(pdf_file_path, dpi=300)

# Configure Tesseract for Telugu and English

tesseract_config = r'--oem 3 --psm 6 -l tel+eng'

# Process each image and save the text

with open(output_text_file_path, 'w', encoding='utf-8') as output_file:

    for i, image in enumerate(images):

        # Extract text from the image using Tesseract

        text = pytesseract.image_to_string(image, config=tesseract_config)

        output_file.write(f"Page {i + 1}\n")

        output_file.write(text)

        output_file.write("\n\n")

print(f'OCR text written to file "{output_text_file_path}"')

# Install required packages

!apt-get update -qq

!apt-get install -y -qq poppler-utils tesseract-ocr-tel

!pip install -q Pillow pytesseract pdf2image

from PIL import Image

import pytesseract

import os

from pdf2image import convert_from_path

# File paths

pdf_file_path = '/content/Images_filename123.pdf'

output_text_file_path = '/content/out.txt'

# Check if PDF exists

if not os.path.exists(pdf_file_path):

    raise FileNotFoundError(f"PDF file {pdf_file_path} not found")

# Convert PDF to images

images = convert_from_path(pdf_file_path, dpi=300)

# Tesseract config for Telugu + English

tesseract_config = '--oem 3 --psm 6 -l tel+eng'

# Extract and write text

with open(output_text_file_path, 'w', encoding='utf-8') as f:

    for i, img in enumerate(images, 1):

        text = pytesseract.image_to_string(img, config=tesseract_config)

        f.write(f"Page {i}\n{text}\n\n")

print(f'OCR text written to file "{output_text_file_path}"')

-------------

referece

what if telugu and english both, only english, only telugu, english and hind, only hind? for Tesseract config for Telugu + English; !apt-get install -y -qq poppler-utils tesseract-ocr-tel tesseract_config = '--oem 3 --psm 6 -l tel+eng'

3/3

ChatGPT said:

Great question. You're working with Tesseract OCR and want to understand how to configure it for different language combinations.

Here's how you should set tesseract_config based on the language needs:

Language Mode	Tesseract Language Code	Example Config
Telugu and English	tel+eng	--oem 3 --psm 6 -l tel+eng
Only English	eng	--oem 3 --psm 6 -l eng
Only Telugu	tel	--oem 3 --psm 6 -l tel
English and Hindi	eng+hin	--oem 3 --psm 6 -l eng+hin
Only Hindi	hin	--oem 3 --psm 6 -l hin

---

Additional Setup (if not installed):

Make sure you install the required language packs:

bash
!apt-get install -y -qq tesseract-ocr-eng tesseract-ocr-tel tesseract-ocr-hin

Or you can install individually like:

bash
!apt-get install -y -qq tesseract-ocr-hin  # For Hindi only

---

Notes:

• --oem 3: Use the default OCR Engine mode (both LSTM and legacy).

• --psm 6: Assume a uniform block of text (change if needed based on layout).

• Language order in -l (like tel+eng or eng+hin) affects script recognition priority slightly.

Do you want to auto-detect the language or restrict it manually?

---------------------