Update app.py (#1)

- Update app.py (ec67c5bb9603d286267ac621b2210701c08ad1ef)

app.py

@@ -4,16 +4,36 @@ import numpy as np
 import pickle
 from PIL import Image
 
-# Load model & label encoder sekali di awal
+# === Load Model dan Label Encoder ===
 model = tf.keras.models.load_model("plantopia.h5")
 with open("label_encoder.pkl", "rb") as f:
     label_encoder = pickle.load(f)
 
-# Daftar label
 class_names = label_encoder.inverse_transform(np.arange(len(label_encoder.classes_)))
 
+# === Optional: Panduan Perawatan ===
+treatment_guides = {
+    "Mildew": "🧴 Gunakan fungisida berbahan aktif sulfur atau kalium bikarbonat.",
+    "Fungal": "🌿 Pangkas daun yang terinfeksi. Gunakan fungisida sistemik.",
+    "Insect": "🐛 Gunakan insektisida nabati seperti neem oil.",
+    "Dead Leaf": "🗑️ Buang daun mati dan berikan nutrisi tambahan.",
+    "Blight": "☠️ Gunakan fungisida berbasis tembaga.",
+    "Rust": "🔶 Gunakan fungisida spesifik karat seperti myclobutanil.",
+    "Spot Diseases": "🎯 Periksa kelembapan, gunakan fungisida daun.",
+    "Healthy": "✅ Tanaman sehat. Lanjutkan perawatan optimal.",
+    "Septoria": "💉 Gunakan fungisida berbasis klorotalonil.",
+    "Virus": "🦠 Cabut tanaman yang terinfeksi berat.",
+    "Early Blight": "🌗 Gunakan fungisida berbahan aktif mankozeb.",
+    "Late Blight": "🌘 Gunakan fungisida sistemik dan buang bagian busuk.",
+    "Bacterial": "🧼 Semprot dengan larutan antibakteri."
+}
+
+# === Fungsi Prediksi ===
 def predict(img: Image.Image):
     try:
+        if img.mode != "RGB":
+            img = img.convert("RGB")
+
         img = img.resize((224, 224))
         img_array = np.array(img) / 255.0
         img_array = np.expand_dims(img_array, axis=0)
@@ -21,26 +41,35 @@ def predict(img: Image.Image):
         prediction = model.predict(img_array)[0]
         predicted_index = np.argmax(prediction)
         predicted_label = class_names[predicted_index]
-        predicted_confidence = prediction[predicted_index] * 100
+        predicted_confidence = float(prediction[predicted_index]) * 100
+        rekomendasi = treatment_guides.get(predicted_label, "⚠️ Belum ada rekomendasi.")
 
-        result = f"✅ Prediksi penyakit tanaman: {predicted_label}\n"
-        result += f"🎯 Akurasi keyakinan model (confidence): {predicted_confidence:.2f}%\n\n"
-        result += "--- Detail Probabilitas untuk Semua Kelas ---\n"
-        for label, prob in zip(class_names, prediction):
-            result += f"{label:15}: {prob:.4f}\n"
-
-        return result
+        return {
+            "output": f"✅ Prediksi: {predicted_label} | 🎯 Confidence: {predicted_confidence:.2f}%",
+            "label": predicted_label,
+            "confidence": round(predicted_confidence, 2),
+            "rekomendasi": rekomendasi,
+            "raw_probs": prediction.tolist(),
+            "class_names": class_names.tolist()
+        }
 
     except Exception as e:
-        return f"❌ Terjadi error saat memproses gambar: {str(e)}"
-
+        return {
+            "output": f"❌ Error: {str(e)}",
+            "label": None,
+            "confidence": 0,
+            "rekomendasi": "❌ Tidak tersedia.",
+            "raw_probs": [],
+            "class_names": []
+        }
 
+# === UI Gradio ===
 gr.Interface(
     fn=predict,
-    inputs=gr.Image(type="pil"),
-    outputs="text",
-    title="Deteksi Penyakit Daun 🌿",
-    description="Upload gambar daun dan sistem akan memprediksi penyakitnya berdasarkan model CNN yang telah dilatih.",
+    inputs=gr.Image(type="pil", label="📸 Upload Gambar Daun"),
+    outputs="json",
+    title="🌿 Plantopia Enterprise API",
+    description="Prediksi penyakit tanaman berbasis gambar, confidence, rekomendasi perawatan, dan distribusi probabilitas. Output dalam format JSON siap pakai.",
     allow_flagging="never",
-    api_name="predict"              
-).launch(show_error=True) 
+    api_name="predict"
+).launch(show_error=True)