dotch3 / harvtech-vegetables-health-ai

The AI project to assess the health of vegetables for HarvTech.

Vegetables Health

The AI project to assess the vegetables health.

Support vegetables:

• Bean;
• Potato;
• Tomato;
• Rice.

About models

We are using models from Kaggle.

Gemini API

You can get a Gemini API key in this link.

Environment setup

Create a .env file in the root folder with content:

CLOUD_STORAGE_URL_PREFIX=https://URL_TO_YOUR_STORAGE_DOMAIN/
FLASK_PORT=5001
GEMINI_API_KEY=YOUR_API_KEY_HERE
FLASK_ENV=production

Prepare

Download the best_tomato_leaf_inceptionV3_256.h5 and best_rice_leaf.h5, put their in constants/weights directory.

Install and download models

Use pip install -U -r requirements.txt to install dependencies.

Use ./download_models.sh to download the trained models

Install certificates

If in development use self-signed certificates, in production follow the steps of your SSL authorizer

Development

Run ./generate_certificates.sh

Start

Run python3 app.py to start project.

The https://API_URL:PORT/predict should be used for the POST API calls

Following an example of the API body request

{
    "model_type": "tomato_leaf",
    "content_url": "https://techwarriors-objectstorage-test.s3.us-south.cloud-object-storage.appdomain.cloud/0034a551-9512-44e5-ba6c-827f85ecc688___RS_Erly.B%209432.png"
}

Tests

Run the directory tests, if in PyCharm .it will execute all files with application running.

Roadmap:

• Unit tests;
• Create a REST API to use this code;
• Kaggle explains;
• Tests describe how to execute;
• Deploy project in VPS;
• Validation URL from back-end with prefix in DNS of cloud (Jorge);
• ChatGPT return with data structured;
• Create all tests with different models;
• Search new tomato models in Kaggle;
• Integrated responses with ChatGPT;
• Upload vegetable images in cloud (Jorge);
• Configure CORS.