🌱 Plant Disease Classifier - Streamlit App

This is a web-based deep learning app built using Streamlit that allows users to detect plant diseases from leaf images. It leverages a trained Convolutional Neural Network (CNN) model on the PlantVillage dataset to identify 38 different plant conditions (including healthy and diseased states).

---

---

Trained Model Link is given below

https://drive.google.com/file/d/1rKh-IElSdHTqax7XdfSdZTn-r8T_qWPf/view?usp=drive_link)https://drive.google.com/file/d/1rKh-IElSdHTqax7XdfSdZTn-r8T_qWPf/view?usp=drive_link

---

---

🔍 Features

• 🌿 Detect plant diseases from uploaded leaf images.
• 🧪 Shows prediction confidence.
• 📊 Displays class probabilities as a horizontal bar chart.
• 💡 Clean dark theme with Orbitron animated UI.
• 📷 Image is resized and previewed before prediction.

---

📁 Project Structure

.
├── app.py                         # Streamlit web app
├── plant_disease_prediction_model.h5  # Trained Keras CNN model
├── class_indices.json             # Label dictionary (class index to class name)
├── requirements.txt               # List of required Python packages
├── README.md                      # This file

---

🔧 Installation

🔹 Clone the repository:

git clone https://github.com/yourusername/plant-disease-classifier.git
cd plant-disease-classifier

🔹 Install the dependencies:

pip install -r requirements.txt

Make sure Python 3.8+ is installed.

---

🚀 Running the App

streamlit run app.py

• Upload a leaf image (JPG or PNG).
• Click 🔍 Predict.
• View the predicted class, confidence, and probability chart.

---

🧠 Model Details

• ✅ Trained on: PlantVillage dataset
• 🔢 Classes: 38 (healthy + diseased conditions of crops)
• 🏗️ Architecture: CNN with 3 Conv2D layers + MaxPooling + Dense layers
• 🎯 Accuracy: ~95% on validation data

---

📚 Example Classes

• Apple___Black_rot
• Grape___Esca_(Black_Measles)
• Tomato___Late_blight
• Potato___healthy
• Pepper,_bell___Bacterial_spot
• ...and 30+ more

---

---

📜 License

MIT License — use freely with attribution.

---