app.py

import streamlit as st
import numpy as np
import json
import os
from PIL import Image
import tensorflow as tf

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

# ── Load model ─────────────────────────────
@st.cache_resource
def load_model():
    model_path = "plant_model.keras"
    indices_path = "class_indices.json"

    # Debug: list files in current directory
    current_dir = os.getcwd()
    files = os.listdir(current_dir)
    debug_files = f"Files in {current_dir}: {files}"

    if not os.path.exists(model_path):
        return None, None, None, f"Model not found: {model_path}. {debug_files}"

    if not os.path.exists(indices_path):
        return None, None, None, f"class_indices.json not found: {indices_path}. {debug_files}"

    try:
        model = tf.keras.models.load_model(model_path, compile=False)
    except Exception as e:
        return None, None, None, f"Failed to load model: {e}. {debug_files}"

    try:
        with open(indices_path, "r") as f:
            raw = json.load(f)
    except Exception as e:
        return None, None, None, f"Failed to load class indices: {e}"

    # Handle both class index formats
    if all(str(k).isdigit() for k in raw.keys()):
        class_indices = {int(k): v for k, v in raw.items()}
    else:
        class_indices = {v: k for k, v in raw.items()}

    input_shape = model.input_shape
    return model, class_indices, input_shape, None


# ── Prediction ─────────────────────────────
def predict(model, class_indices, img, input_shape):
    # Use model's expected input size
    if input_shape and len(input_shape) >= 3:
        target_size = (input_shape[1], input_shape[2])  # (height, width)
    else:
        target_size = (160, 160)  # fallback

    img = img.convert("RGB").resize(target_size)
    arr = np.array(img, dtype=np.float32) / 255.0
    arr = np.expand_dims(arr, axis=0)

    preds = model.predict(arr, verbose=0)[0]

    top5_idx = np.argsort(preds)[::-1][:5]
    top5 = [(class_indices.get(i, "Unknown"), float(preds[i]) * 100) for i in top5_idx]

    return top5[0][0], top5[0][1], top5


# ── UI ─────────────────────────────
st.title("🌿 LeafScan - Plant Disease Detector")

# Verify XSRF is disabled (optional - can remove after confirming)
st.write("XSRF Protection:", st.get_option("server.enableXsrfProtection"))

model, class_indices, input_shape, error = load_model()

if error:
    st.error(error)
    st.stop()

st.success(f"Model loaded — {len(class_indices)} classes")
st.info(f"Model input shape: {input_shape}")

uploaded = st.file_uploader("Upload a leaf image", type=["jpg", "png", "jpeg"])

if uploaded:
    img = Image.open(uploaded)
    st.image(img, caption="Uploaded Image", use_column_width=True)

    if st.button("Analyze"):
        try:
            with st.spinner("Predicting..."):
                class_name, conf, top5 = predict(model, class_indices, img, input_shape)

            plant, disease = class_name.split("___") if "___" in class_name else ("Unknown", class_name)

            st.subheader("Result")
            st.write("🌱 Plant:", plant.replace("_", " "))
            st.write("🦠 Disease:", disease.replace("_", " "))
            st.write(f"🎯 Confidence: {conf:.2f}%")

            st.subheader("Top Predictions")
            for name, p in top5:
                st.write(f"{name}: {p:.2f}%")
        except Exception as e:
            st.error(f"Prediction failed: {e}")