💚

Why We Built LeafSense

Food security starts with healthy plants

Plant diseases threaten global food security, causing billions of dollars in crop losses annually. Early and accurate detection is crucial for effective treatment, but expert pathologists aren't always available — especially in rural communities.

LeafSense bridges this gap by putting the power of deep learning into a simple web interface. Upload a photo of a plant leaf, and our AI model identifies the plant species and diagnoses potential diseases in seconds — providing actionable treatment recommendations.

14 Plant Species
26 Diseases Detected
~99.5% Accuracy
Free Forever
✨

What Makes LeafSense Special

Designed with real-world needs in mind

🔒

Privacy First

Images are processed on the server and deleted immediately. No data is stored or shared with third parties.

🌍

Open Source

Fully open-source codebase. View, contribute, and learn from the code on GitHub. Built for the community.

💚

Completely Free

No fees, no subscriptions, no hidden costs. Accessible to everyone - from students to small-scale farmers.

🧠

Explainable AI

Transparent predictions with top-3 results, confidence scores, and detailed disease information for each diagnosis.

🔧

Technology Stack

Built with modern, industry-standard tools

🔧 Backend

Python 3.10+ Flask Gunicorn Pillow

🧠 Machine Learning

PyTorch TorchVision ResNet50 Transfer Learning

🎨 Frontend

HTML5 CSS3 JavaScript Jinja2

📂 Data & Training

PlantVillage Kaggle Jupyter Notebooks Matplotlib

🚀 Deployment

Hugging Face Gunicorn Git GitHub
đŸŽ¯

Project Objectives

Deep Learning project milestones

1

Dataset Collection & Preparation

Source the PlantVillage Dataset from Kaggle and preprocess 54,305 images across 38 classes.

2

Custom CNN Development

Design and train a baseline CNN architecture from scratch to establish performance benchmarks.

3

Transfer Learning (ResNet50)

Implement ResNet50 with custom classifier head, achieving ~99.5% accuracy through fine-tuning.

4

Model Evaluation & Comparison

Compare Custom CNN vs ResNet50 using accuracy, confusion matrices, and classification reports.

5

Application Interface

Build a responsive, multi-page Flask web application with intuitive drag-and-drop upload interface.

6

Agricultural Knowledge Base

Curate disease information including severity levels, treatment plans, and prevention strategies.

7

Backend Integration & Deployment

Integrate trained model with Flask backend and prepare for production deployment.

👨‍đŸ’ģ

Meet the Developer

The mind behind LeafSense

Developer Avatar

Developer & Creator

ÂĨ@$# Kakadiya

Machine Learning Engineer & Full-Stack Developer

Passionate about leveraging technology to solve real-world problems.LeafSense was born from the intersection of desire to learn deep learning and a commitment to make agricultural insights accessible to everyone.

âš ī¸

Important Disclaimer

LeafSense is a hobby project developed for educational and research purposes. It is not a substitute for professional agricultural consultation. While the model achieves high accuracy on the test dataset, real-world conditions (lighting, camera quality, leaf angle) may affect results. Always consult a qualified agronomist or plant pathologist for definitive diagnosis and treatment decisions.

Try LeafSense Now

Upload a leaf image and experience AI-powered plant disease detection

đŸ”Ŧ Start Diagnosis