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 to OpenCV
        img = pil_img.convert("RGB")
        img = np.array(img)
        img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)

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

        # Thresholding to highlight digits
        _, binary = cv2.threshold(gray, 120, 255, cv2.THRESH_BINARY_INV)

        # Resize for better OCR
        resized = cv2.resize(binary, None, fx=2, fy=2, interpolation=cv2.INTER_CUBIC)

        # Run OCR with digit whitelist
        config = r'--oem 3 --psm 6 -c tessedit_char_whitelist=0123456789.'
        raw_text = pytesseract.image_to_string(resized, config=config)

        print("🔍 OCR Raw Output:", repr(raw_text))  # Show in Hugging Face logs

        # Filter for digits only
        weight = ''.join(filter(lambda c: c in '0123456789.', raw_text))
        confidence = 95 if weight else 0
        return weight.strip(), confidence

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