src/app.py

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

st.set_page_config(page_title="Plant Disease Expert System", layout="wide", page_icon="🌿")

# PATH DEFINISI - Sesuaikan dengan folder src/model di repo Anda
BASE_MODEL_DIR = os.path.join("src", "model")
PATH_JSON = os.path.join(BASE_MODEL_DIR, "knowledge_base.json")

MODEL_PATHS = {
    "MobileNetV2 (Best Fine-Tuned)": "model_skripsi_mobilenet_final_FT.h5",
    "InceptionV3 (Fine-Tuned)": "model_skripsi_inception_finetuned.h5",
    "Deep Custom CNN": "model_skripsi_cnn_deep.h5",
    "MobileNetV2 (Standard)": "model_skripsi_mobilenet.h5",
    "InceptionV3 (Standard)": "model_skripsi_inception.h5"
}

# Fungsi memuat model secara dinamis (Hanya 1 model di RAM)
@st.cache_resource
def load_selected_model(model_key):
    try:
        path = os.path.join(BASE_MODEL_DIR, MODEL_PATHS[model_key])
        if os.path.exists(path):
            return tf.keras.models.load_model(path)
        st.error(f"File model tidak ditemukan: {path}")
        return None
    except Exception as e:
        st.error(f"Gagal memuat model: {e}")
        return None

@st.cache_data
def load_labels(path):
    if os.path.exists(path):
        with open(path, 'r') as f:
            return json.load(f)
    return {}

# UI Utama
st.sidebar.header("⚙️ Kontrol Sistem")
model_choice = st.sidebar.selectbox("Pilih Arsitektur Model:", list(MODEL_PATHS.keys()))

KB_DATA = load_labels(PATH_JSON)
model = load_selected_model(model_choice)

if model and KB_DATA:
    st.title("🌿 Identifikasi Penyakit Tanaman")
    uploaded_file = st.file_uploader("Unggah foto daun...", type=["jpg", "jpeg", "png"])

    if uploaded_file:
        img = Image.open(uploaded_file).convert("RGB")
        st.image(img, caption="Preview Foto", width=400)
        
        if st.button("🚀 Jalankan Diagnosa"):
            # Preprocessing otomatis sesuai model
            h, w = model.input_shape[1], model.input_shape[2]
            img_res = img.resize((h, w))
            img_arr = np.expand_dims(np.array(img_res) / 255.0, axis=0)

            preds = model.predict(img_arr)
            idx = np.argmax(preds)
            conf = np.max(preds) * 100

            res = KB_DATA.get(str(idx))
            if res:
                st.success(f"### {res['jenis']} ({conf:.2f}%)")
                st.write(f"**Saran:** {res.get('deskripsi', 'N/A')}")