Plant Disease Classification using PyTorch, FastAPI

This project aims to detect plant diseases using deep learning techniques implemented with PyTorch. It provides both a web application built with Flask for real-time detection and a user-friendly interface powered by Streamlit for offline analysis.

Dataset

The dataset used in this project is available on Hugging Face: Plant Disease Dataset

Docker Image

You can find the Docker image containing the complete working environment and application at Docker Hub: Plant Disease Detection Docker Image

Features

• Utilizes PyTorch and torchvision for training and deploying deep learning models.
• Provides a Flask API for real-time inference.
• Offers a Streamlit web application for offline analysis and visualization.
• Supports a wide range of plant diseases for accurate detection.

Project Structure

• model: Directory containing trained models.
• src: Directory containing the source code.
  • Models: Model dir
    • resnet.py: Implementation of ResNET from scratch.
  • datasets: Directory for deep learning model scripts.
    • plant_disease.py: script for creating custom dataset.
  • helper.py: script which contains helperfunctions.
  • train.py: script for training loop and class.
• .gitignore: File specifying ignored files and directories for version control.
• README.md: This README file.
• dockerfile: Dockerfile for building Docker image.
• main.ipynb: Jupyter notebook containing main code or experiments.
• main.py: Main Python script for running the application.
• requirements.txt: File specifying project dependencies.

Installation

1. Clone this repository:

   git clone https://github.com/khushwant04/Plant-Disease.git
   

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Streamlit app:

   uvicorn app:app --reload
   

Testing images:

Download Testing Images.

https://1drv.ms/f/s!Akr767JWN3vEllsH0PqUESUpbakN?e=rETLSX