Potato Disease Detection System

AI-Powered Plant Health Assessment

Author

Tam Le

Published

October 2, 2025

Project Overview

This is a simple machine learning project that demonstrates a potato disease detection system. The project includes:

A CNN model trained on the PlantVillage dataset
A FastAPI backend that serves the model
A React frontend for uploading images
Documentation created with Quarto

🎬 Watch Demo Video

See the complete application in action - from image upload to disease prediction results.

What This Project Shows

This project demonstrates my ability to: - Build end-to-end ML applications - Create REST APIs with FastAPI - Develop React frontends - Document projects professionally with Quarto - Work with TensorFlow/Keras

System Architecture

Code

flowchart TD
    A[Upload Image] --> B[React Frontend]
    B --> C[FastAPI Backend]
    C --> D[TensorFlow Model]
    D --> E[Classification Result]
    E --> B

Technology Stack

Backend: - FastAPI - TensorFlow - PIL (Pillow) - NumPy

Frontend: - React - Material-UI - Axios - Material-UI Dropzone

Model: - CNN trained on PlantVillage dataset - 3 classes: Early Blight, Late Blight, Healthy

Project Structure

/api - FastAPI backend with model serving
/frontend - React application
/models - Trained TensorFlow model
/training - Jupyter notebook with training code
Quarto documentation (this website)