# Palm Oil Ripeness AI - Mobile App

This Flutter application ports the AI-driven palm oil ripeness detection system to mobile devices. It allows for gallery-based analysis with local history storage and health alerts.

## 🛠 Features
- **AI Inference**: On-device TFLite processing for ripeness classification.
- **Gallery Analysis**: Pick photos directly from the device gallery.
- **History Vault**: Local storage of detection results using SQFlite.
- **Health Alerts**: Immediate UI feedback for abnormal bunches.

## 🚀 Setup & Installation

### Prerequisites
- [Flutter SDK](https://docs.flutter.dev/get-started/install) installed and added to your PATH.
- Android Studio or VS Code with Flutter extension.
- A physical Android device connected via USB with "Developer Options" and "USB Debugging" enabled.

### 1. Initialize Project
Navigate to the mobile project directory and fetch dependencies:
```bash
cd palm_oil_mobile
flutter pub get
```

### 2. Run the App
To test the app with full AI performance, run it in **Release Mode**:
```bash
flutter run --release
```

## 🧪 Testing Instructions

1.  **Launch the App**: Open the app on your device.
2.  **Select Gallery Analysis**: Tap on the **Analyze Gallery** card.
3.  **Pick a Photo**: Grant gallery permissions and select a photo of a palm oil bunch.
4.  **Observe Results**:
    *   The app will display "Initializing AI..." while processing.
    *   The detected grade and confidence score will appear.
    *   🔴 **Health Alert**: If "Empty_Bunch" or "Abnormal" is detected, a red warning box will show.
5.  **Check History**: Navigate back to the Home screen and select **History Vault** to see your past detections.
6.  **Re-examine**: Tap on any history card to view the detailed summary.

## 🏗 Project Structure (Flutter Beginner's Guide)

As you're new to Flutter, here’s a map of where the important logic lives. All your custom code is inside the `lib/` folder:

### 1. `lib/main.dart`
The **entry point** of the application. This is where the app theme is defined and the first screen (`HomeScreen`) is launched.

### 2. `lib/models/`
*   **`palm_record.dart`**: Defines the "blueprint" for a detection result. If you want to store more data (like GPS coordinates or weather), you'd add fields here.

### 3. `lib/services/`
This folder contains "behind-the-scenes" logic:
*   **`yolo_service.dart`**: Wraps the AI model. This is where the TFLite model is initialized and where image inference happens.
*   **`database_helper.dart`**: Manages the local SQLite database (`palm_oil_history.db`). Change this if you need to update the database schema.

### 4. `lib/screens/`
This is where the **UI** lives:
*   **`home_screen.dart`**: The main dashboard.
*   **`analysis_screen.dart`**: The UI for image selection, showing the loading spinner, and displaying the AI results.
*   **`history_screen.dart`**: The list view for previous records and the detail modal.

### 5. `assets/`
Located at the root of the project. This contains your non-code files:
*   `models/best.tflite`: The trained brain of the app.
*   `labels/labels.txt`: The names of the classes the model can recognize.

## ⚙️ Configuration
*   **`pubspec.yaml`**: The project's manifest. This is where you add new dependencies (like camera or GPS plugins) and register new assets.