hashinikarunathilaka/gemstone-classifier-api

Gemstone Classifier API

A RESTful API for classifying gemstone images using a pre-trained deep learning model.

Features

• Image classification for 85 different types of gemstones
• High accuracy prediction with confidence scores
• RESTful API with proper error handling
• Swagger UI documentation
• Containerized deployment with Docker

API Endpoints

The API is versioned with a prefix: /api/v1

Root Endpoint

GET /

Returns basic information about the API and available endpoints.

Health Check

GET /api/v1/health

Checks if the API is running and the model is loaded.

List Gemstone Classes

GET /api/v1/classes

Returns a list of all gemstone classes that the model can predict.

Predict Gemstone

POST /api/v1/predict

Predicts the gemstone class from an uploaded image.

Request:

• Content-Type: multipart/form-data
• Body: file (image file)

Response:

{
  "prediction": "Ruby",
  "confidence": 0.98,
  "timestamp": "2023-04-01T12:34:56.789Z"
}

API Documentation

The API documentation is available at /docs when the server is running.

Installation and Setup

Prerequisites

• Python 3.10 or higher
• TensorFlow 2.15.0
• FastAPI 0.109.0

Local Development

1. Clone the repository
2. Install dependencies:

   pip install -r requirements.txt
   

3. Run the server:

   uvicorn app:app --reload
   

4. Access the API at http://localhost:8000
5. Access the API documentation at http://localhost:8000/docs

Docker Deployment

1. Build the Docker image:

   docker build -t gemstone-classifier-api .
   

2. Run the container:

   docker run -p 7860:7860 gemstone-classifier-api
   

3. Access the API at http://localhost:7860
4. Access the API documentation at http://localhost:7860/docs

Environment Variables

The following environment variables can be set to configure the API:

• MODEL_PATH: Path to the model file (default: gemstone_classifier.h5)
• CLASS_NAMES_PATH: Path to the class names file (default: class_names.csv)

Model Information

The model is a convolutional neural network trained on a dataset of gemstone images. It can classify 85 different types of gemstones with high accuracy.

Input image requirements:

• The model expects images to be resized to 150x150 pixels
• Images can be in any common format (JPEG, PNG, etc.)
• Images should be clear and well-lit for best results

License

This project is licensed under the MIT License - see the LICENSE file for details.