Plant Disease Classification

Introduction

This project leverages the power of EfficientNet model to help farmers and gardeners quickly and accurately identify plant diseases using images of plant leaves. Upon detecting a disease, the application provides information about the disease, including its name, cause, and symptoms.

Project Overview

1. Training Pipeline
2. Build a simple server
3. Build client

Dataset

Data is collected from various sources on Kaggle and aggregated together. The dataset has 22 plants with corresponding diseases splitting into train, val, and test.

Data Description

Name	Disease
Apple	alternate leaf spot, brown spot, gray spot, healthy leaf,
Bell pepper	bacterial spot, healthy
Cassava	bacterial blight, brown streak, green mottle, healthy mosaic
Cherry	healthy, powdery mildew
Chili	healthy, leaf curl, leaf spot, whitefly, yellowish
Citrus	black spot, canker, greening, healthy, melanoma
Coconut	caterpillars, drying of leaflets, flaccidity, leaflets, yellowing
Coffee	healthy, red spider mite, rust
Corn	common rust, gray leaf spot, healthy
Grape	black rot, blight, esca, healthy
Guava	canker, dot, healthy, mummification, rust
Jack fruit	algal spot, black spot, healthy
Mango	anthracnose, back die, bacterial canker, cutting weevil, gall midge, healthy, mildew powder mould sooty,
Peach	bacterial spot, healthy,
Potato	early blight, healthy, late blight
Rice	blast, blight, brown spot, healthy, narrow brown spot, scald
Soybean	bacterial blight, caterpillar, diabrotica speciosa, downy mildew, healthy, mosaic virus, powdery mildew, rust, southern blight
Strawberry	healthy, leaf scorch
Sugarcane	healthy, red rot, red stripe, rust,
Tea	bird eye spot, brown blight, healthy, leaf spot
Tomato	bacterial spot, curl virus, early blight, healthy, late blight, leaf mold, mosaic virus, septoria leaf spot, spot
Wheat	brown rust, healthy, yellow rust

Set-up

• Clone this repo to your Local Machine and move into it.

$ git clone [email protected]:tinh2044/PlantDisease_classification.git
$ cd PlantDisease_classification

• Create virtual environments with conda to avoid conflicts

conda create --name plantDisease
conda activate plantDisease

• Install requirements.

pip install -r ./requirements.txt

Training

Download dataset from kaggle link and extract its to Datasets folder

Or download datasets using Kaggle CLI

kaggle datasets download -d nguyenchitinh/plantdisease-with-20-plant

Make sure after extracted, your Datasets folder has struct like this

|——Datasets
   |——AppleDisease
        |——train
       |——class_name_1
       |——class_name_2
       ......
       |——class_name_n
  |——valid
       |——class_name_1
       |——class_name_2
       ......
       |——class_name_n
  |—-test
       |——class_name_1
       |——class_name_2
       ......
       |——class_name_n
   |——BellPepperDisease
       ......

• To train the models run

python train_multiple_model.py --epoch 100 --batch_size 32 --root_dir ./Datasets --img_size 224 --export_dir ./SavedModels --h5_dir ./Models

After training is complete. Weights of model is saved to ./Model and SavedModel is saved to ./SavedModels

• Evaluate the model

python evaluate.py --root_dir ./Datasets --h5_dir ./Models

• Convert model to tflite

python covert_tflite.py

Server

Make sure you have to copy all tflite model in ModelLight to server/ModelLight

Move to server directory

cd server

• Run server by command

uvicorn app.main:app --host 127.0.0.1 --port 5000

• Or using docker

docker compose up

Client

Move to client directory

cd client

Run client by command

npm start

Note: You need to create .env file in client folder and type

REACT_APP_API_URL=your_server_api

References

• New Plant Diseases Dataset
• Plant Disease Classification Merged Dataset
• EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
• Transfer learning and fine-tuning