MAIZA API- Backed API for Maize Leaf Disease Prediction

MAIZA API is a backend project that classifies the health status of a Maize Leaf from captured Image.

The models are trained on data from [Mensah et al](#maiza-api--backed-api-for-maize-leaf-disease-prediction)
The model can classify the image into seven classes, ie., Fallarmyworm, Grasshopper, Healthy, Leafbeetle, Leafblight, Leafspot, and Streakvirus.

API Link

Endpoints available:

• / : Welcome page (GET)
• /status: Returns the status of the API (GET)
• /predict: Returns the prediction and recommendation (POST)

The endpoint for prediction and classification is at '/predict'. The endpoint only accepts POST requests.

Required data and field labels for the prediction are:

• latitude: latitude coordinate of the user
• longitude: longitude coordinate of the user
• image: uploaded image file of the maize file

Below is a demonstration of a POST request and the returned response.

Requirements

• python 3.10 +
• docker (optional)

Installation

• Clone this repository to your local machine and change into the directory

  git clone https://github.com/hawkinswinja/maiza-api.git && cd maiza-api
  

• Download the models.tar file and extract in the root of the project folder

  tar -xf models.tar
  

• Create and activate a virtual environment

  python -m venv virt && source virt/bin/activate
  
  # windows
  python -m venv virt && source virt\Scripts\activate
  

• Install dependencies listed in requirements.txt

  python -m pip install -r requirements.txt
  

• launch the server using gunicorn (ubuntu)

  gunicorn server.app:app
  

  • use waitress for windows

  waitress-serve --listen=*:8000 server.app:app
  

• access the url at 127.0.0.1:8000

alternatively if using docker, run the following command to run the image

docker pull hawkinswinja/maiza:v4
docker run -p 8000:5000 -v ./server/.env:/app/.env 

© hawkinswinja