🌿 Plant Disease Detector

---

📌 Overview

🌱 Plant Disease Detector is a deep-learning web app that classifies leaf images into healthy or diseased categories using a custom Convolutional Neural Network (CNN). It leverages data augmentation, batch normalization, and dropout for robust performance, then wraps the trained model in a Flask interface for real-time predictions.

---

🔍 Key Features

• 🖼️ Image Classification: Detects multiple plant diseases from leaf photos
• 🎛️ Data Augmentation: Rotation, shifts, shear, zoom & flips for better generalization
• 🚀 Fast Inference: Lightweight Flask app serving predictions in seconds
• 📊 Visual Insights: Training/validation loss & accuracy plots for analysis
• 🔄 Persistence: Save & load models with pickle for reuse

---

🛠️ Tech Stack

Component	Library / Tool
Model Building	Keras (TensorFlow)
Image Processing	OpenCV, NumPy
Web Framework	Flask
Data Handling	scikit-learn, pickle
Visualization	Matplotlib

---

📂 Project Structure

plant-disease-detector/
├── archive/                 # Dataset root
│   └── PlantVillage/
├── static/
│   └── uploads/             # Uploaded images for Flask
├── templates/
│   └── index.html           # Flask UI template
├── Output Graphs/           # 📊 Plots for model performance (accuracy/loss)
│   ├── Output (1).jpg
│   ├── Output (2).jpg
│   ├── Output (3).jpg
│   ├── Output (4).jpg
│   ├── Output (5).jpg
│   ├── Output (6).jpg
│   ├── Output (7).jpg
│   ├── Output (8).jpg
│   └── Output (9).jpg
├── main.ipynb                # Jupyter notebook for training & export
├── app.py                    # Flask application
├── mod.py                    # Prediction helper functions (check function)
├── cnn_model.pkl             # Trained model
├── label_transform.pkl       # Label binarizer
├── requirements.txt          # Python dependencies
└── README.md

---

🔗 Link to Dataset

Download the PlantVillage dataset from Kaggle:

📥 PlantVillage Dataset on Kaggle

---

🚀 Deployment

1. Download the Dataset

   • Clone or download the PlantVillage dataset into archive/PlantVillage.

2. Install Dependencies

   pip install -r requirements.txt
   
   

3. Train & Export Model

   • Run main.ipynb notebook to train the CNN model.
   • This will generate cnn_model.pkl and label_transform.pkl.

4. Run the App

   python app.py

5. Access the Web Interface

   • Navigate to http://127.0.0.1:5000 in your browser.

---

📊 Output Visualizations

1. Training and Validation Accuracy

2. Training and Validation Loss

---

🎯 Model Performance

• Training Accuracy: ~98%
• Validation Accuracy: ~96%
• Test Accuracy: ~95% (may vary based on dataset split)

---

🤝 Authors

• @Avnish1447

---

License

This project is licensed under the MIT License — see the LICENSE file for details.

Happy diagnosing! 🌾🔬😁