README.md

metadata

title: AgriVision AI
emoji: 🌿
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 5.34.2
python_version: '3.10'
app_file: app.py
pinned: false

🌿 AgriVision AI — Plant Disease Detection using Deep Learning

AgriVision AI is an end-to-end Deep Learning + Computer Vision project that detects plant diseases from leaf images using Transfer Learning with EfficientNetB0.

The system predicts 38 different plant disease categories with high accuracy and provides:

• Disease Prediction
• Confidence Scores
• Grad-CAM Visualization
• Disease Description
• Symptoms
• Prevention Methods
• Cure Suggestions

Built using TensorFlow, OpenCV, EfficientNet, and Gradio.

---

🚀 Features

✅ Plant Disease Detection using AI ✅ 38 Disease Categories ✅ Transfer Learning with EfficientNetB0 ✅ Fine-Tuned Deep Learning Model ✅ Leaf Segmentation for Better Predictions ✅ Grad-CAM Explainability Visualization ✅ Top-3 Predictions with Confidence Bars ✅ Disease Information & Cure Suggestions ✅ Interactive Gradio Web Application ✅ Real-Time Image Prediction System

---

🧠 Problem Statement

Plant diseases significantly reduce agricultural productivity and crop quality.

Traditional disease identification:

• requires expert knowledge
• is time-consuming
• may delay treatment

AgriVision AI helps farmers and researchers instantly identify plant diseases using leaf images.

---

🌱 Why This Project Matters

This project combines:

• Artificial Intelligence
• Agriculture
• Deep Learning
• Computer Vision
• Explainable AI

Applications:

• Smart Farming
• Precision Agriculture
• AI-Based Crop Monitoring
• Agricultural Decision Support Systems

---

🏗️ Project Workflow

Leaf Image
   ↓
Leaf Segmentation
   ↓
Image Preprocessing
   ↓
EfficientNetB0 Model
   ↓
Disease Prediction
   ↓
Grad-CAM Visualization
   ↓
Disease Information & Cure Suggestions

---

🛠️ Tech Stack

Programming Language

• Python

Deep Learning Frameworks

• TensorFlow
• Keras

Computer Vision

• OpenCV
• Grad-CAM

Deployment

• Gradio

Image Processing

• NumPy
• Pillow
• rembg

---

📂 Dataset

PlantVillage Dataset

• 54,000+ Images
• 38 Classes
• Multiple Crops & Diseases

Dataset Source: https://www.kaggle.com/datasets/emmarex/plantdisease

---

📊 Disease Categories

The model supports 38 classes including:

• Apple Diseases
• Corn Diseases
• Tomato Diseases
• Potato Diseases
• Grape Diseases
• Strawberry Diseases
• Peach Diseases
• Pepper Diseases
• Soybean Diseases
• Healthy Leaf Detection

---

🧪 Model Development Phases

Phase 1 — Dataset Preparation

Performed:

• Data Loading
• Train/Validation Split
• Image Augmentation

Techniques:

• Rotation
• Zoom
• Horizontal Flip
• Rescaling

---

Phase 2 — Baseline CNN Model

Built a custom CNN using:

• Conv2D
• MaxPooling
• BatchNormalization
• Dropout
• Dense Layers

Purpose:

• Establish baseline performance
• Understand CNN workflow

---

Phase 3 — Transfer Learning

Used:

EfficientNetB0

Advantages:

• Better Feature Extraction
• Higher Accuracy
• Fewer Parameters
• Faster Training

Initially froze pretrained layers.

---

Phase 4 — Fine Tuning

Unfroze upper EfficientNet layers and retrained using:

• Low Learning Rate
• Additional Epochs

Result: ✅ Significant performance improvement

---

📈 Final Results

Metric	Value
Validation Accuracy	96%
Model	EfficientNetB0
Classes	38
Dataset Size	54K+ Images
Framework	TensorFlow/Keras

---

🔍 Explainable AI with Grad-CAM

Grad-CAM highlights regions of the image influencing model predictions.

Benefits:

• Improves transparency
• Helps visualize model focus
• Makes predictions explainable

---

✂️ Leaf Segmentation

Implemented background removal using:

rembg

Benefits:

• Reduces background noise
• Improves internet image prediction
• Better real-world generalization

---

🌐 Web Application

Built using:

Gradio

Features:

• Upload Leaf Image
• View Top Predictions
• Confidence Bars
• Grad-CAM Heatmap
• Disease Information
• Cure Suggestions

---

📁 Project Structure

Plant_Disease/
│
├── app.py
├── README.md
├── requirements.txt
├── .gitignore
│
├── utils/
│   ├── predict.py
│   ├── gradcam.py
│   ├── segmentation.py
│   └── disease_info.py
│
├── sample_images/
│
└── notebooks/

---

⚙️ Installation

1. Clone Repository

git clone YOUR_GITHUB_REPO_LINK
cd Plant_Disease

---

2. Create Virtual Environment

python -m venv venv

---

3. Activate Environment

Windows

venv\Scripts\activate

Mac/Linux

source venv/bin/activate

---

4. Install Dependencies

pip install -r requirements.txt

---

📥 Download Trained Model

Due to GitHub file size limitations, the trained model is hosted externally.

Download Model Here: _Click_HERE

After downloading, place the model inside:

Plant_Disease/
│
├── final_agrivision_model.keras
├── app.py
└── utils/

---

▶️ Run Application

python app.py

Application runs at:

http://127.0.0.1:7860

---

📦 Required Libraries

tensorflow
opencv-python
gradio
numpy
pillow
matplotlib
rembg
onnxruntime

---

🧠 Key Learnings

Through this project I learned:

• Transfer Learning
• EfficientNet Architecture
• CNN Fundamentals
• Fine Tuning
• Grad-CAM Explainability
• Leaf Segmentation
• Deep Learning Deployment
• Real-World Image Challenges
• Domain Shift Problems
• Model Generalization

---

⚠️ Real-World Challenges

The model performs strongly on PlantVillage-style images.

Challenges with internet images include:

• Complex backgrounds
• Different lighting conditions
• Blurry images
• Domain shift
• Real-world variability

Future improvements:

• Real farm dataset training
• Lesion segmentation
• Higher resolution models
• Advanced explainability techniques

---

🚀 Future Improvements

• Mobile App Deployment
• Multi-Language Support
• Real-Time Webcam Detection
• Cloud Deployment
• Disease Severity Estimation
• PDF Report Generation
• Advanced Explainability Methods
• Farmer Advisory System

---

👨‍💻 Author

Mohd Faizanullah

AI/ML Enthusiast | Deep Learning | Computer Vision | Generative AI

---

⭐ Support

If you like this project, give it a star ⭐ on GitHub.