🌿 Plant Village Classification

Author: Subarna Saikia

Dataset:

---

This project classifies plant leaf images into 15 distinct categories using a deep learning model built on MobileNetV2. The dataset is taken from the PlantVillage collection.

🚀 Features

• Uses Transfer Learning with MobileNetV2
• Data Augmentation for better generalization
• Early Stopping, Model Checkpointing & Learning Rate Scheduling
• Achieves over 93% validation accuracy

🧠 Classes

15 classes including bacterial spot, early/late blight, mold, viruses, and healthy leaves across pepper, potato, and tomato.

🛠️ Model Architecture

• Base: MobileNetV2 (frozen)
• Custom Dense layers: 1024 → 768 → 512 → Softmax(15)

📈 Training Highlights

• Epochs: 25
• Accuracy: ~95% (training), ~93% (validation)

🔗 Notebook

Check the full code in main.ipynb.

📦 Requirements

TensorFlow
NumPy
OpenCV
Matplotlib