Farmer Eye Robotic Car – Graduation Project

Smart AI & IoT System for Real-Time Plant Disease Detection

---

Project Overview

Farmer Eye is a professional-grade, end-to-end AI-IoT ecosystem designed to modernize agriculture by automating plant disease diagnostics. The system integrates a remote-controlled robotic car, high-performance Deep Learning models, and a Raspberry Pi-powered edge device to patrol fields and identify crop diseases in real-time.

By bridging the gap between hardware and software, Farmer Eye provides farmers with instant diagnostic feedback and localized treatment recommendations (available in English and Arabic) to prevent crop loss and optimize harvest health.

Mobile Application

The system includes a dedicated cross-platform mobile application built with Flutter, serving as the central hub for monitoring and control:

• 🎥 Real-Time Live Feed: Low-latency video streaming from the robotic car's onboard camera.
• 🔔 Instant Alerts: Push notifications sent the moment a plant disease is detected, including classification and confidence metrics.
• 💊 Treatment Intelligence: Integrated pharmaceutical database providing clinical diagnostics and treatment protocols.
• 🕹️ Remote Telemetry: Real-time status monitoring for hardware health and connectivity.

System Features

• Real-Time Edge Inference: Continuous monitoring and detection powered by localized processing on Raspberry Pi.
• High-Accuracy CNN: Fine-tuned Convolutional Neural Networks optimized for high-precision identification across various plant classes.
• Multi-Crop Support: Robust detection for Cotton, Tomato, Potato, Pepper, and Strawberry.
• Bi-Lingual Diagnostics: Comprehensive treatment guidance in both English and Arabic.
• Production-Ready Architecture: Decoupled, modular codebase designed for scalability and maintainability.
• IoT-Cloud Synchronization: WebSocket-based communication ensuring instant data delivery between edge and mobile.

Tech Stack

Artificial Intelligence & Data

• Frameworks: TensorFlow, Keras, Scikit-learn
• Libraries: NumPy, Pandas, OpenCV, PIL (Pillow)
• Deep Learning: Convolutional Neural Networks (CNN)

Hardware & IoT

• Compute: Raspberry Pi
• Camera: PiCamera2 / HD Modules
• Mechanics: Robotic Car Chassis, L298N Motor Drivers
• Connectivity: WebSockets (Asyncio)

Mobile & Frontend

• Framework: Flutter (Dart)
• State Management: Provider / BLoC
• Communication: WebSocket Client

Backend & Infrastructure

• Server: Python-based WebSocket Server
• Database: Excel/CSV-based treatment reference (Openpyxl)

Project Structure

FarmerEye/
├── data/
│   └── plant_disease_data.xlsx      # Database for treatments and diagnostics
├── docs/
│   └── assets/
│       └── robotic_car_image.jpg    # Project visual assets
├── models/
│   ├── fine_tuned_model.h5          # Optimized CNN model
│   └── plant_disease_model_final.h5 # Final production-ready model
├── notebooks/
│   └── research_and_training.ipynb  # ML development and training pipeline
├── src/
│   ├── app.py                       # Main application entry point
│   ├── combined_detection_stream.py # Combined UI and streaming logic
│   ├── real_time_detection.py       # Core inference and hardware logic
│   └── raspberry_pi_camera_stream.py# Low-level camera streaming service
├── tests/                           # System validation and testing
├── requirements.txt                 # Dependency manifest
└── README.md                        # Project documentation

Installation

1. Clone the Repository

   git clone https://github.com/mariiammaysara/FarmerEye.git
   cd FarmerEye

2. Environment Setup

   python -m venv venv
   source venv/bin/activate  # Windows: venv\Scripts\activate

3. Install Dependencies

   pip install -r requirements.txt

Hardware Setup

1. Camera Configuration:
   • Enable the camera interface on Raspberry Pi.
   • Install Picamera2 according to the official Raspberry Pi documentation.
2. Motor Driver:
   • Connect the motor driver to the GPIO pins as configured in the source code.
3. Power Management:
   • Ensure stable power supply for both the Raspberry Pi and the motor chassis.

Usage

1. Training & Research

Explore the model development phase via Jupyter:

jupyter notebook notebooks/

2. Real-Time Detection

Start the monitoring system on the Raspberry Pi:

python src/real_time_detection.py

3. Unified Stream Analysis

Run the combined detection and streaming service:

python src/combined_detection_stream.py

Output

Upon detection, the system provides:

• Disease Classification: Accurate identification of the plant condition.
• Confidence Score: Statistical probability of the detection.
• Treatment Protocol: Actionable advice in English/Arabic fetched from the database.

---

Developed and Designed by
Mariam Maysara • Fatma Zayed • Mohamed Magdy • Mohamed Hesham

FarmerEye Team