app.py

import streamlit as st
import cv2
import numpy as np
import base64
from PIL import Image
import io

def convert_cv2_to_base64(cv2_image):
    """Convert a cv2 image to a base64 string."""
    _, buffer = cv2.imencode('.jpg', cv2_image)
    return base64.b64encode(buffer).decode('utf-8')

def convert_base64_to_cv2(base64_string):
    """Convert base64 string to a cv2 image."""
    try:
        image_bytes = base64.b64decode(base64_string)
        pil_image = Image.open(io.BytesIO(image_bytes))
        cv2_image = np.array(pil_image)
        cv2_image = cv2.cvtColor(cv2_image, cv2.COLOR_RGB2BGR)
        return cv2_image
    except Exception as e:
        print(f"Error decoding base64: {e}")
        return None

def extract_numbers(cv2_image):
    """
    Extracts numbers from the image using OpenCV.
    """
    # Preprocessing
    gray = cv2.cvtColor(cv2_image, cv2.COLOR_BGR2GRAY)
    thresh = cv2.threshold(equalized, 128, 255, cv2.THRESH_BINARY_INV)[1] #invert

    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

    number_contours = []
    for contour in contours:
        x, y, w, h = cv2.boundingRect(contour)
        area = cv2.contourArea(contour)
        if 20 < area < 500 and 10 < w < 100 and 10 < h < 100 and 0.1 < w / h < 1.0: #adjusted
            number_contours.append(contour)

    numbers = []
    for i, contour in enumerate(number_contours):
        x, y, w, h = cv2.boundingRect(contour)
        if x > cv2_image.shape[1] / 4 and x < cv2_image.shape[1]/2 and y < cv2_image.shape[0] / 2:
            number = 1
        elif x > cv2_image.shape[1] / 2 and x < cv2_image.shape[1] * 3/4 and y < cv2_image.shape[0]/2:
            number = 2
        elif x > cv2_image.shape[1] / 4 and x < cv2_image.shape[1]/2 and y > cv2_image.shape[0]/2:
            number = 3
        elif x > cv2_image.shape[1] / 2 and x < cv2_image.shape[1] * 3/4 and y > cv2_image.shape[0]/2:
            number = 4
        else:
            number = i + 1
        numbers.append(number)
    return numbers



def find_regions(cv2_image, number):
    """Find pixels connected to the given number."""
    gray = cv2.cvtColor(cv2_image, cv2.COLOR_BGR2GRAY)
    thresh = cv2.threshold(gray, 128, 255, cv2.THRESH_BINARY)[1]
    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

    region_pixels = []
    min_x, min_y = cv2_image.shape[1], cv2_image.shape[0]
    max_x, max_y = 0, 0
    found_number = False

    for contour in contours:
        x, y, w, h = cv2.boundingRect(contour)
        if 10 < w < 50 and 10 < h < 50 and 0.2 < w / h < 1.0:
            extracted_number = i+1
            if extracted_number == number:
                found_number = True
                for i in range(y, y + h):
                    for j in range(x, x + w):
                        if 0 <= i < cv2_image.shape[0] and 0 <= j < cv2_image.shape[1]:
                            region_pixels.append((j, i))
                            min_x = min(min_x, j)
                            min_y = min(min_y, i)
                            max_x = max(max_x, j)
                            maxY = max(max_y, i)
    if not found_number:
        return [], None
    return region_pixels, (min_x,min_y,max_x,max_y)

def process_and_display(image_file):
    """Process the image and display the result."""
    try:
        # Read image file
        image_bytes = image_file.read()
        pil_image = Image.open(io.BytesIO(image_bytes))
        cv2_image = np.array(pil_image)
        cv2_image = cv2.cvtColor(cv2_image, cv2.COLOR_RGB2BGR)

        # Process the image with OpenCV
        numbers = extract_numbers(cv2_image) #use the function
        st.session_state.numbers = numbers

        # Display
        image_base64 = convert_cv2_to_base64(cv2_image)
        st.image(f"data:image/jpeg;base64,{image_base64}", caption="Uploaded Image", use_container_width=True)

        # Display the numbers
        st.write("Recognized Numbers:", numbers) # <--- ADD THIS LINE

        # Create number buttons
        cols = st.columns(max(1, len(numbers)))  # Ensure at least 1 column is created
        for i, number in enumerate(numbers):
            with cols[i]:
                if st.button(f"Highlight {number}", key=f"highlight_{number}"): # ADDED KEY
                    st.session_state.selected_number = number
                    region, border = find_regions(cv2_image, number)
                    highlighted_image = cv2_image.copy()
                    if border:
                        cv2.rectangle(highlighted_image, (int(border[0]), int(border[1])), (int(border[2]), int(border[3])), (0, 0, 0), 2)
                    for x, y in region:
                         highlighted_image[y, x] = [255, 255, 0]
                    highlighted_image_base64 = convert_cv2_to_base64(highlighted_image)
                    st.image(f"data:image/jpeg;base64,{highlighted_image_base64}", caption=f"Highlighted {number}", use_container_width=True)
    except Exception as e:
        st.error(f"Error processing image: {e}")

def main():
    st.title("Paint by Numbers Solver")
    # st.set_option('deprecation.showfileuploaderlabel', False) # Remove this line
    image_file = st.file_uploader("Upload a Paint by Numbers image", type=["png", "jpg", "jpeg"])

    if image_file is not None:
        # Pass the image file to process_and_display
        process_and_display(image_file) #  CHANGE HERE
        


if __name__ == "__main__":
    main()