import streamlit as st
import numpy as np
import cv2
from keras.models import load_model
from keras.preprocessing.image import img_to_array
from PIL import Image

# Unique Page config
st.set_page_config(
    page_title="VisionGuard: AI Mask Monitor",
    layout="wide"
)

# Load model with caching
@st.cache_resource
def load_model_cached():
    return load_model("project_face_mask_detection.keras")

model = load_model_cached()

# Haar Cascade for face detection
face_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + "haarcascade_frontalface_default.xml")

# Sidebar with your branding
with st.sidebar:
    st.title("VisionGuard: Mask Monitor")
    st.markdown("""
    **VisionGuard** is a real-time system powered by **Convolutional Neural Networks (CNNs)** to help promote public health by detecting face mask compliance.

    **How it works:**
    - Upload or capture an image
    - A CNN-based model detects face(s) and checks for mask presence
    - Results include confidence levels and visual feedback

    This tool was built as a deep learning project to demonstrate practical applications of CNNs in image classification tasks.

    **Built by:** Thirupathirao • 2025
    """)
    st.info("For best results, use clear, front-facing images with good lighting.")
    st.caption("Empowering safety through deep learning.")


# Resize image
def resize_image(image, max_size=(400, 400)):
    image = image.copy()
    image.thumbnail(max_size)
    return image

# Prediction logic
def detect_and_predict(image_input):
    image_np = np.array(image_input.convert("RGB"))
    gray = cv2.cvtColor(image_np, cv2.COLOR_RGB2GRAY)
    faces = face_cascade.detectMultiScale(gray, 1.1, 4)

    if len(faces) == 0:
        return image_input, None, "No face detected"

    x, y, w, h = faces[0]
    face_roi = image_np[y:y+h, x:x+w]
    face_pil = Image.fromarray(face_roi).resize((200, 200))
    img_array = img_to_array(face_pil) / 255.0
    img_array = np.expand_dims(img_array, axis=0)

    prediction = model.predict(img_array)[0][0]
    confidence = (1 - prediction) if prediction < 0.5 else prediction
    label = "Mask Detected" if prediction < 0.5 else "No Mask Detected"

    # Draw results
    color = (0, 255, 0) if prediction < 0.5 else (255, 0, 0)
    cv2.rectangle(image_np, (x, y), (x + w, y + h), color, 2)
    cv2.putText(image_np, f"{label} ({confidence*100:.2f}%)", (x, y - 10),
                cv2.FONT_HERSHEY_SIMPLEX, 0.6, color, 2)

    return Image.fromarray(image_np), confidence, label

# App header
st.markdown("<h1 style='text-align: center;'>VisionGuard: AI Mask Monitor</h1>", unsafe_allow_html=True)
st.markdown("<p style='text-align: center;'>A deep learning-powered tool for mask detection and safety awareness.</p>", unsafe_allow_html=True)

# Input method
input_choice = st.selectbox("Choose Input Method", ["Upload Image", "Use Webcam"])

# Upload option
if input_choice == "Upload Image":
    uploaded_file = st.file_uploader("Upload an image (JPG, JPEG, PNG)", type=["jpg", "jpeg", "png"])

    if uploaded_file:
        image_input = Image.open(uploaded_file)
        resized_display = resize_image(image_input)

        col1, col2 = st.columns(2)

        with col1:
            st.image(resized_display, caption="Uploaded Image")

        with st.spinner("Analyzing with VisionGuard..."):
            result_img, confidence, label = detect_and_predict(image_input)

        with col2:
            st.image(result_img, caption="Detection Output")
            if confidence is not None:
                st.metric("Confidence Score", f"{confidence*100:.2f}%")
                if "Mask" in label:
                    st.success("✔️ You're following safety measures!")
                else:
                    st.error("⚠️ No mask detected! Please wear a mask in public spaces.")
            else:
                st.warning(label)

# Webcam option
elif input_choice == "Use Webcam":
    col1, col2 = st.columns([1, 3])

    with col1:
        camera_image = st.camera_input("Take a picture using your webcam")

    if camera_image:
        image_input = Image.open(camera_image)

        with st.spinner("Analyzing..."):
            result_img, confidence, label = detect_and_predict(image_input)

        with col2:
            st.write("Analysis Result")
            st.image(result_img, caption="Detection Output")
            if confidence is not None:
                st.metric("Confidence Score", f"{confidence*100:.2f}%")
                if "Mask" in label:
                    st.error("No mask detected! Please wear a mask in public spaces.")
                else:
                    st.success("You're following safety measures!")

            else:
                st.warning(label)