utils.py

import easyocr
import cv2
import matplotlib.pyplot as plt
from PIL import Image
from style_classifier import classify_style
from style_model import load_style_model

import easyocr
import cv2
import numpy as np
from style_classifier import classify_style
from style_model import load_style_model

def process_document(image_path, style_model):
    """
    Uses EasyOCR to detect & recognize text bounding boxes and classifies text style.
    """
    reader = easyocr.Reader(['en'], gpu=False)
    ocr_results = reader.readtext(image_path, detail=1)  # detail=1 => (bbox, text, conf)
    
    original_bgr = cv2.imread(image_path)
    final_output = []
    
    for (bbox, text, conf) in ocr_results:
        (x_min, y_min), (_, _), (x_max, y_max), (_, _) = bbox
        cropped_bgr = original_bgr[int(y_min):int(y_max), int(x_min):int(x_max)]  # Convert to integer indices
        
        style_label, style_conf = classify_style(cropped_bgr, style_model)
        
        final_output.append({
            "text": text,
            "ocr_conf": conf,
            "bbox": bbox,
            "style": style_label,
            "style_conf": style_conf
        })
    
    return final_output


def display_image(image_path):
    """ Helper function to display the selected image """
    img_pil = Image.open(image_path)
    plt.imshow(img_pil)
    plt.title(f"Selected Image: {image_path}")
    plt.axis("off")
    plt.show()