NST & YOLO API

API combining YOLO Object Segmentation and Neural Style Transfer

Description

This API allows you to perform object segmentation using YOLOv8 and apply neural style transfer to images in 3 different modes. The main goal is to segment selected classes and apply style to the class/ these classes or whole image. Execution time is 0.5s - 1.3s per call and it can vary depending on the server.

Endpoints

Get YOLO Model classes

Endpoint: /get-classes

Method: GET

Description: Retrieves the list of YOLO model classes available for object segmentation

Response:

{
   {
    "0": "person",
    "1": "bicycle",
    "2": "car",
    "3": "motorcycle",
    "4": "airplane",
    ...
    }
}

Get Neural Style Transfer trained models

Endpoint: /get-styles

Method: GET

Description: Retrieves the list of style models that can be applied for Neural Style Transfer.

Response:

{
    [
    {
        "author": "Eugène Delacroix",
        "description": "",
        "model_id": "",
        "title": "Liberty Leading the People",
        "year": 1830
    },
    {
        "author": "Salvador Dali",
        "description": "",
        "model_id": "",
        "title": "The Persistence of Memory",
        "year": 1931
    },
    {
        "author": "Petar Lubarda",
        "description": "",
        "model_id": "",
        "title": "Crni Panter",
        "year": 1968
    },
    {
        "author": "Petar Lubarda",
        "description": "",
        "model_id": "",
        "title": "Konji",
        "year": 1953
    },
    {
        "author": "Uros Predic",
        "description": "",
        "model_id": "",
        "title": "Vesela Braca",
        "year": 1887
    },
    {
        "author": "Dragos Kalajic",
        "description": "",
        "model_id": "",
        "title": "Sokolarka",
        "year": 2001
    },
    ...
]
}

Process Image

Endpoint: /process-image

Method: POST

Description: Applies object segmentation and neural style transfer to the provided image.

Request:

You can use example request listed in the repository by name send_request.py.

• image: The image file to be processed.
• info: Additional information containing the following fields:
  • style: The model_id value of selected style to be applied for Neural Style Transfer
  • mode: The mode for applying the style. Possible values: "CLASS_MODE", "INVERSE_MODE", "FULL_STYLE_MODE".

Response:

{
  "execution_time": "1.42",
  "processed_image_path": "http://localhost:5000/static/processed_images/b1daf3ac-3585-4aaa-a292-27bd7cc33926.jpg"
}

The API will respond with the link to modified image based on the applied object segmentation and neural style transfer.

How to run API

to be done..

Usage

To use the API, make requests to the provided endpoints using the appropriate HTTP methods.

Example usage with cURL: to be done..

Attribution

add the attribution

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Contact

For any questions or inquiries, please contact [email protected]