ocr_engine.py

import cv2
import pytesseract
from PIL import Image
import re

def extract_weight(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    
    # OCR
    text = pytesseract.image_to_string(gray, config='--psm 6')

    # Clean and lower text
    text = text.lower().replace('\n', ' ').strip()

    # Regex to find pattern like 52.25 g or 75.8 kg
    match = re.search(r'(\d+\.\d+|\d+)\s*(kg|g)', text)

    if match:
        number = match.group(1)
        unit = match.group(2)
        return f"{number} {unit}"
    else:
        return "Weight not detected"