Plant Disease Prediction Using Deep Learning

Webapp -------------------------------------------------------------------------------------------------- git Repo

• The project aims to develop a system that can accurately identify various diseases in plants using machine learning algorithms.

• The system will be trained on a dataset of images of plants affected by different diseases and will use deep learning techniques to extract relevant features from the images.

• The extracted features will then be used to train a model that can classify new images of plants based on the presence or absence of disease.

• The project will involve the use of deep learning frameworks such as TensorFlow, Keras, and will require the use of image processing techniques such as convolutional neural networks (CNNs) and transfer learning.

• Other technologies that are used to built this project are React JS, Streamlit,javascript, FastAPI,Google Cloud Platforms.

• The ultimate goal of the project is to provide a tool that can help farmersand plant pathologists quickly identify and treat plant diseases, leading to higher crop yields and better food security.

Dataset

The PlantVillage dataset is a collection of over 50,000 high-resolution images of 26 different plant species and 38 different plant diseases, created for advancing computer vision and machine learning techniques for plant disease diagnosis and classification.

Model architecture

• This project uses a transfer learning approach to tackle our image classification task, leveraging the pre-trained VGG16 model as our base.

• I added my own custom layers on top of the VGG16 architecture to fine-tune the model to my specific problem domain.

• This allowed me to benefit from the powerful feature extraction capabilities of VGG16 while also tailoring the model to my unique data and classification needs.

• The architecture is as followed :

    1. Resizing layer
    2. VGG16 layers (only Cnn layers)
    3. Flattening layer
    4. A dense layer with 128 neurons and relu function
    5. A dense layer with 64 neurons and relu function
    6. Finally an output layer with Softmax function.
  

Results

The Performance of the CNN model is as followed :

Installation

Install my project on your device locally.

1. clone the entire Repository.

git clone https://github.com/suraj4502/DL_project_Plant_Disease_prediction/

2. go the api folder and install all the requirements from requirements.txt file and execute the main.py .

cd api
pip install -r requirements.txt
python main.py

3. go to frontend folder and install the requirements and run the reacts Js app.

cd frontend
npm install --from-lock-json
npm audit fix
npm run start

4. To run the Streamlit App.

cd Streamlit_app
pip install -r requirements.txt
Streamlit run 🏠Home.py

Demo

1. Streamlit APP

The app is deployed on the Streamlit Cloud while the model is deployed on the GCP as a cloud function, a request is sent every time we want to predict the class of the plant and a response is received from GCP.

• Login Page:

• Signing up:

• Retrieving Credentials:

• Main section:

• Uploading an image to get result:

• Taking an image from camera to get results :

• Disease Information Page:

2. React JS App :

Credits

This project was created by Surajkumar Yadav, contact at [email protected] for any queries.

---