Plant Leaf Disease Detection using Deep Learning

Plant Leaf Disease Detection using Deep Learning, Transfer Learning & Gradio

I built an end-to-end Computer Vision system that detects plant leaf diseases from images using Deep Learning and Transfer Learning techniques.

This project covers the full ML pipeline: dataset preparation, model design, evaluation, comparison between multiple CNN architectures, and deployment through an interactive web interface.

🧠 Models Implemented

I designed and compared three different architectures:

Simple CNN

Deep CNN

MobileNetV2 (Transfer Learning with ImageNet)

The best model was automatically selected based on F1-Score after evaluating all models on a separate test set.

📊 Model Performance (Test Accuracy)

Simple CNN:

Deep CNN:

MobileNetV2: ✅ Best Model

Models were evaluated using:

Accuracy

Precision

Recall

F1-Score

⚙️ Tech Stack & Tools

TensorFlow / Keras

NumPy, Pandas, Matplotlib

Scikit-learn (metrics)

tf.data pipeline

Transfer Learning (MobileNetV2)

SplitFolders for dataset splitting

Gradio for model deployment

Mendeley Plant Leaf Diseases Dataset

💡 What this project demonstrates

✅ CNN architecture design from scratch

✅ Transfer Learning in real practice

✅ Proper model evaluation beyond accuracy

✅ Efficient image pipeline using TensorFlow

✅ Deploying a DL model as a usable web app

The final system allows users to upload a leaf image in the browser and get the top predicted diseases with probabilities in real time.

This project connects Deep Learning + Evaluation + Deployment into one complete AI application.