Source Code

import subprocess
from typing import List
from doctr.io import DocumentFile
from doctr.io.elements import Page
from doctr.models import ocr_predictor
from doctr.models.predictor.pytorch import OCRPredictor
from numpy import ndarray
import pytesseract
import re
import string
from time import perf_counter


def apply_text_filter(func):
    def wrapper(self, *args, **kwargs):
        text = func(self, *args, **kwargs)
        return self.text_filter(text)
    return wrapper


class OCRBase:
    def __init__(self):
        pass

    def process(self):
        pass

    def text_filter(self, text):
        return re.sub(r'(?m)^\s*\S{1,3}\s*$\n?', '', text).strip()


class OCRDoctr(OCRBase):
    """
    python3.12 -m pip install doctr
    """
    def __init__(self):
        print('Loading OCR Model...')
        self.model: OCRPredictor = ocr_predictor(det_arch='db_resnet50', reco_arch='crnn_vgg16_bn', pretrained=True)
        print('Finished')

    @apply_text_filter
    def process(self, image_path):
        doc: List[ndarray] = DocumentFile.from_images(image_path)
        result: Page = self.model(doc)
        return result.render()


class OCRTerreract(OCRBase):
    """
    sudo apt update
    sudo apt install tesseract-ocr
    sudo apt install libtesseract-dev
    
    python3.12 -m pip install pytesseract
    """
    def __init__(self):
        whitelist = string.printable.replace('"', '\\"').replace("'", "\\'")
        self.config = fr'-c tessedit_char_whitelist=" {whitelist}"'
        pass

    @apply_text_filter
    def process(self, image_path):
        text = pytesseract.image_to_string(image_path, config=self.config, timeout=20)
        return text

class OCRRobertKnight(OCRBase):
    """
    curl https://sh.rustup.rs -sSf | sh
    sudo apt install cargo
    cargo install ocrs-cli
    """
    def __init__(self):
        pass

    @apply_text_filter
    def process(self, image_path):
        text = subprocess.run(['ocrs', image_path], capture_output=True, text=True).stdout
        return text


if __name__ == "__main__":
    file_paths = {
        '/home/USER/Documents/images/images/1719103056310.png': 'its over',
        '/home/USER/Documents/images/images/halt.png': 'grinch',
        '/home/USER/Documents/images/images/th-1384675618.jpeg': 'heaven',
        '/home/USER/Documents/images/images/toontown.jpg': 'towntown',
        '/home/USER/Documents/images/images/max_boost.png': 'max_boost',
        '/home/USER/Documents/images/images/ireland.jpg': 'ireland',
    }

    d = {
        'tesseract': OCRTerreract(),
        'doctr': OCRDoctr(),
        'ocrs': OCRRobertKnight(),
    }
    
    for file_path, file_alias in file_paths.items():
        print()
        print('---')
        print()
        for name, obj in d.items():
            start = perf_counter()
            text = obj.process(file_path)
            elapsed_time = perf_counter() - start

            print('- ![](url){style="max-width: 500px;"}')
            print(f'### Library: {name}')
            print()
            print(f'**Time:** {elapsed_time:.3f} seconds')
            print()
            print(f'```text\n{text}\n```')