Update app.py

app.py

@@ -3,119 +3,126 @@ from ultralytics import YOLO
 import pandas as pd
 import datetime
 import os
-import cv2
-from PIL import Image
+import numpy as np
+import matplotlib.pyplot as plt
 
-# 1. KONFIGURASI MODEL & DATABASE
+# 1. KONFIGURASI
 MODEL_PATH = 'yolov8n.pt'
 DB_FILE = 'inventory_log.csv'
-
-# Inisialisasi Model
 model = YOLO(MODEL_PATH)
 
-# 2. FUNGSI DATABASE
-def save_to_database(df_results):
-    """Menyimpan hasil deteksi ke file CSV lokal di server Hugging Face"""
-    if df_results.empty:
-        return "Tidak ada data untuk disimpan."
+# 2. FUNGSI FORECASTING (PREDIKSI)
+def predict_demand():
+    if not os.path.isfile(DB_FILE):
+        return "⚠️ Data histori belum cukup untuk prediksi.", None
+    
+    df = pd.read_csv(DB_FILE)
+    if len(df) < 3: # Butuh minimal 3 data point untuk melihat tren
+        return "⚠️ Butuh minimal 3 kali scan untuk menghitung tren prediksi.", None
+    
+    # Kelompokkan data berdasarkan barang dan hitung rata-rata jumlah per hari
+    df['Timestamp'] = pd.to_datetime(df['Timestamp'])
+    df['Date_Ordinal'] = df['Timestamp'].apply(lambda x: x.toordinal())
+    
+    forecast_results = "### 🔮 Prediksi Ketersediaan Stok:\n"
+    items = df['Barang'].unique()
     
-    file_exists = os.path.isfile(DB_FILE)
+    fig, ax = plt.subplots(figsize=(8, 4))
     
-    # Tambahkan Timestamp untuk setiap baris data
-    df_results['Timestamp'] = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    for item in items:
+        item_df = df[df['Barang'] == item].sort_values('Timestamp')
+        X = item_df['Date_Ordinal'].values.reshape(-1, 1)
+        y = item_df['Jumlah'].values
+        
+        # Linear Regression Sederhana (Slope & Intercept)
+        if len(y) > 1:
+            slope, intercept = np.polyfit(X.flatten(), y, 1)
+            
+            # Jika tren menurun (slope negatif)
+            if slope < 0:
+                days_left = int(-intercept / slope) - datetime.date.today().toordinal()
+                days_left = max(0, days_left)
+                forecast_results += f"- **{item}**: Diperkirakan habis dalam **{days_left} hari**.\n"
+            else:
+                forecast_results += f"- **{item}**: Stok cenderung stabil/meningkat.\n"
+            
+            # Plotting histori
+            ax.plot(item_df['Timestamp'], item_df['Jumlah'], marker='o', label=f"Tren {item}")
+
+    ax.set_title("Grafik Perubahan Stok Barang")
+    ax.set_ylabel("Jumlah Unit")
+    ax.legend()
+    plt.xticks(rotation=45)
+    plt.tight_layout()
     
-    # Simpan/Append ke CSV
-    df_results.to_csv(DB_FILE, mode='a', index=False, header=not file_exists)
-    return f"✅ Database diperbarui: {datetime.datetime.now().strftime('%H:%M:%S')}"
+    return forecast_results, fig
 
-# 3. FUNGSI UTAMA (DETEKSI & PROSES)
+# 3. FUNGSI UTAMA (INTEGRASI SCAN)
 def process_inventory(img):
     if img is None:
-        return None, None, "⚠️ Silakan unggah foto.", None
+        return None, None, "⚠️ Unggah foto.", None, "Silakan scan dulu.", None
 
-    # --- Tahap 1: Deteksi AI ---
     results = model(img)
-    res_plotted = results[0].plot() # Gambar hasil deteksi
+    res_plotted = results[0].plot()
     
-    # --- Tahap 2: Ekstraksi Data ---
     detections = results[0].boxes.cls.tolist()
     names = model.names
     counts = {}
-    
     for class_id in detections:
         name = names[int(class_id)]
         counts[name] = counts.get(name, 0) + 1
     
-    # --- Tahap 3: Persiapan Tabel & Logika Bisnis ---
     inventory_list = []
-    if counts:
-        for item, count in counts.items():
-            # Simulasi Threshold: Jika di bawah 5, tandai sebagai Low Stock
-            status = "✅ STOK AMAN" if count >= 5 else "🚨 LOW STOCK"
-            inventory_list.append({
-                "Barang": item.upper(),
-                "Jumlah": count,
-                "Status": status
-            })
+    for item, count in counts.items():
+        status = "✅ AMAN" if count >= 5 else "🚨 LOW STOCK"
+        inventory_list.append({"Barang": item.upper(), "Jumlah": count, "Status": status, "Timestamp": datetime.datetime.now()})
     
     df_inventory = pd.DataFrame(inventory_list)
     
-    # --- Tahap 4: Update Database ---
-    db_status = save_to_database(df_inventory.copy())
+    # Save to Database
+    if not df_inventory.empty:
+        file_exists = os.path.isfile(DB_FILE)
+        df_inventory.to_csv(DB_FILE, mode='a', index=False, header=not file_exists)
     
-    # Jika tabel kosong
-    if df_inventory.empty:
-        return res_plotted, None, "ℹ️ Tidak ada objek logistik terdeteksi.", None
-
-    # Return: Gambar, Tabel, Pesan Status, dan Path File CSV untuk diunduh
-    return res_plotted, df_inventory, db_status, DB_FILE
+    # Jalankan Forecasting setelah scan
+    forecast_text, forecast_plot = predict_demand()
+    
+    return res_plotted, df_inventory, "✅ Database Updated!", DB_FILE, forecast_text, forecast_plot
 
-# 4. ANTARMUKA GRADIO (DASHBOARD UI)
+# 4. UI GRADIO (TABBED INTERFACE)
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("""
-    <div style="text-align: center;">
-        <h1>🏙️ IntelliStock AI: Warehouse Management System</h1>
-        <p>Solusi <b>Computer Vision</b> untuk otomasi inventarisasi dan monitoring stok real-time.</p>
-    </div>
-    """)
+    gr.Markdown("# 🏙️ IntelliStock AI: End-to-End Warehouse Intelligence")
     
-    with gr.Row():
-        # Kolom Kiri: Kontrol & Input
-        with gr.Column(scale=1):
-            gr.Markdown("### 📸 Input Visual")
-            input_img = gr.Image(type="numpy", label="Upload Foto Gudang/Rak")
-            
-            with gr.Row():
-                clear_btn = gr.Button("🗑️ Clear", variant="secondary")
-                scan_btn = gr.Button("🔍 Run AI Scan", variant="primary")
-            
-            gr.Markdown("---")
-            gr.Markdown("### 📥 Laporan Database")
-            db_msg = gr.Markdown("*Belum ada aktivitas scan*")
-            output_file = gr.File(label="Download Full Inventory Log (CSV)")
+    with gr.Tab("📸 Scan & Monitoring"):
+        with gr.Row():
+            with gr.Column():
+                input_img = gr.Image(type="numpy", label="Input Camera/Upload")
+                scan_btn = gr.Button("🔍 Run AI Scan & Update", variant="primary")
+            with gr.Column():
+                output_img = gr.Image(label="Visual Detection")
+                output_table = gr.Dataframe(label="Current Inventory")
+        
+        with gr.Row():
+            output_file = gr.File(label="📥 Download Database (CSV)")
+            db_status = gr.Markdown()
 
-        # Kolom Kanan: Hasil AI
-        with gr.Column(scale=2):
-            gr.Markdown("### 🕵️ AI Detection Result")
-            output_img = gr.Image(label="Annotated Warehouse View")
-            
-            gr.Markdown("### 📈 Tabel Inventaris Terdeteksi")
-            output_table = gr.Dataframe(
-                headers=["Barang", "Jumlah", "Status"],
-                label="Current Scan Report",
-                interactive=False
-            )
+    with gr.Tab("🔮 Demand Forecasting"):
+        gr.Markdown("### Analisis Prediksi Stok Berdasarkan Histori")
+        with gr.Row():
+            with gr.Column():
+                forecast_output_text = gr.Markdown("Lakukan minimal 3x scan untuk melihat prediksi.")
+            with gr.Column():
+                forecast_output_plot = gr.Plot(label="Grafik Tren Stok")
+        refresh_btn = gr.Button("🔄 Refresh Analisis Prediksi")
 
-    # Definisi Interaksi
+    # Click Events
     scan_btn.click(
         fn=process_inventory,
         inputs=input_img,
-        outputs=[output_img, output_table, db_msg, output_file]
+        outputs=[output_img, output_table, db_status, output_file, forecast_output_text, forecast_output_plot]
     )
     
-    clear_btn.click(lambda: (None, None, "*Aktivitas dibersihkan*", None), 
-                   outputs=[input_img, output_img, db_msg, output_file])
+    refresh_btn.click(fn=predict_demand, outputs=[forecast_output_text, forecast_output_plot])
 
-# 5. RUN APP
 if __name__ == "__main__":
     demo.launch()