Streamlit: AI Web Apps 🌐

import streamlit as st

# Layout
st.title("My AI App")
st.header("Section heading")
st.subheader("Sub-heading")
st.write("Any text, dataframe, chart, dict — all rendered automatically")
st.markdown("**Bold**, *italic*, and `code` formatting supported")

# Input widgets
name = st.text_input("Your name", placeholder="e.g. Priya")
age  = st.number_input("Age", min_value=5, max_value=100, value=15)
lang = st.selectbox("Preferred language", ["English", "Telugu", "Hindi", "Malayalam"])
conf = st.slider("Confidence threshold", min_value=0.0, max_value=1.0, value=0.5)
clicked = st.button("Run prediction")

# File upload
uploaded = st.file_uploader("Upload image", type=["jpg","jpeg","png"])

# Output
st.success("Great job! 🎉")    # green box
st.warning("Check your input")  # yellow box
st.error("Invalid file")        # red box

# Show image
if uploaded:
    st.image(uploaded, caption="Uploaded image", use_container_width=True)

# Show data and charts
import pandas as pd
df = pd.DataFrame({"Score": [85, 92, 78]}, index=["Run1","Run2","Run3"])
st.dataframe(df)
st.bar_chart(df)

# plant_disease_app.py
# Run locally: streamlit run plant_disease_app.py
# Deploy free: https://share.streamlit.io

import streamlit as st
import numpy as np
from PIL import Image

# ── Page config ──
st.set_page_config(
    page_title="Plant Disease Detector",
    page_icon="🌿",
    layout="centered"
)

# ── Cache the model — loaded once, not on every re-run ──
@st.cache_resource
def load_model():
    """Load the trained CNN model. Cached so it only loads once."""
    try:
        import tensorflow as tf
        model = tf.keras.models.load_model("plant_disease_model.h5")
        return model
    except Exception:
        return None

# ── Class labels (matches training dataset order) ──
CLASSES = [
    "Cashew Anthracnose",
    "Cashew Gumosis",
    "Cassava Bacterial Blight",
    "Cassava Brown Spot",
    "Coconut Leaf Blight",
    "Coconut Leaf Rot",
    "Healthy",
    "Maize Leaf Blight",
    "Maize Leaf Spot",
    "Tomato Early Blight"
]

TREATMENTS = {
    "Cashew Anthracnose":      "Spray Mancozeb 75WP (2.5g/L). Remove infected branches.",
    "Cashew Gumosis":          "Apply Bordeaux paste on affected areas. Improve drainage.",
    "Cassava Bacterial Blight":"Remove and burn infected parts. Use disease-free cuttings.",
    "Cassava Brown Spot":      "Apply Carbendazim 50WP (1g/L). Ensure good spacing.",
    "Coconut Leaf Blight":     "Spray Copper oxychloride (3g/L). Improve air circulation.",
    "Coconut Leaf Rot":        "Apply Bordeaux mixture. Remove fallen debris.",
    "Healthy":                 "No treatment needed. Plant is healthy! ✅",
    "Maize Leaf Blight":       "Spray Propiconazole (1ml/L). Use resistant varieties.",
    "Maize Leaf Spot":         "Apply Zineb 75WP (2g/L). Destroy infected crop residue.",
    "Tomato Early Blight":     "Spray Chlorothalonil 75WP (2g/L). Stake plants well."
}

# ── App UI ──
st.title("🌿 Plant Disease Detector")
st.markdown("Upload a photo of a plant leaf to identify disease and get treatment advice.")
st.markdown("---")

model = load_model()
if model is None:
    st.warning("⚠️ Model file not found. Running in demo mode with random predictions.")

# Sidebar — settings
with st.sidebar:
    st.header("Settings")
    confidence_threshold = st.slider("Minimum confidence", 0.3, 1.0, 0.6)
    show_top3 = st.checkbox("Show top 3 predictions", value=True)
    st.markdown("---")
    st.markdown("**Supported crops:**")
    st.markdown("Cashew · Cassava · Coconut · Maize · Tomato")

# File upload
uploaded_file = st.file_uploader(
    "📷 Upload a leaf image",
    type=["jpg", "jpeg", "png"],
    help="Take a clear photo of the leaf in good lighting"
)

if uploaded_file is not None:
    # Display uploaded image
    image = Image.open(uploaded_file).convert("RGB")
    col1, col2 = st.columns([1, 1])
    with col1:
        st.image(image, caption="Uploaded leaf", use_container_width=True)

    with col2:
        with st.spinner("Analysing..."):
            # Preprocess for CNN (224×224, normalised)
            img_array = np.array(image.resize((224, 224))) / 255.0
            img_array = np.expand_dims(img_array, axis=0)  # (1, 224, 224, 3)

            if model is not None:
                predictions = model.predict(img_array)[0]
            else:
                # Demo mode: fake predictions for illustration
                predictions = np.random.dirichlet(np.ones(len(CLASSES)) * 0.5)

            top_idx  = np.argmax(predictions)
            top_conf = predictions[top_idx]
            top_cls  = CLASSES[top_idx]

        if top_conf >= confidence_threshold:
            if top_cls == "Healthy":
                st.success(f"✅ **{top_cls}**\n\n{top_conf*100:.1f}% confident")
            else:
                st.error(f"⚠️ **{top_cls}**\n\n{top_conf*100:.1f}% confident")
        else:
            st.warning(f"Low confidence ({top_conf*100:.1f}%). Please upload a clearer photo.")

    # Treatment recommendation
    if top_conf >= confidence_threshold:
        st.markdown("### 💊 Recommended Treatment")
        treatment = TREATMENTS.get(top_cls, "Consult an agricultural officer.")
        st.info(treatment)

    # Top 3 predictions bar chart
    if show_top3:
        st.markdown("### 📊 Top 3 Predictions")
        top3_idx   = np.argsort(predictions)[-3:][::-1]
        top3_data  = {CLASSES[i]: float(predictions[i]) for i in top3_idx}
        import pandas as pd
        st.bar_chart(pd.Series(top3_data))

    # Feedback button
    st.markdown("---")
    correct = st.radio("Was this prediction correct?", ["Yes", "No"], horizontal=True)
    if st.button("Submit feedback"):
        st.success("Thank you! Your feedback helps improve the model.")