ocr_engine.py

import cv2
import pytesseract
import numpy as np
from PIL import Image

def extract_weight_from_image(pil_img):
    try:
        # Convert PIL image to OpenCV format
        img = pil_img.convert("RGB")
        img_np = np.array(img)
        img_cv = cv2.cvtColor(img_np, cv2.COLOR_RGB2BGR)

        # Convert to grayscale
        gray = cv2.cvtColor(img_cv, cv2.COLOR_BGR2GRAY)

        # Adaptive Thresholding for 7-segment LCD
        processed = cv2.adaptiveThreshold(
            gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY_INV, 15, 10
        )

        # Resize to enhance small text
        resized = cv2.resize(processed, None, fx=2, fy=2, interpolation=cv2.INTER_LINEAR)

        # OCR config tuned for digit blocks
        config = r'--oem 3 --psm 7 -c tessedit_char_whitelist=0123456789.'

        # Run OCR
        text = pytesseract.image_to_string(resized, config=config)

        print("🔍 RAW OCR OUTPUT:", repr(text))

        # Clean the text
        weight = ''.join(c for c in text if c in '0123456789.')
        weight = weight.strip()

        confidence = 95 if weight else 0
        return weight, confidence

    except Exception as e:
        print("❌ OCR Error:", str(e))
        return "", 0