limin-vision 0.2.0

A Python library for working with vision models.

limin-vision

A Python library for working with vision models.

Installation

Install the library using pip:

pip install limin-vision

Usage

After you've installed the library, you can use it by importing the limin_vision module and calling the functions you need. You will also need to provide an API key for your API either by running export OPENAI_API_KEY=$YOUR_API_KEY or by creating an .env file in the root directory of your project and adding the following line:

OPENAI_API_KEY=$YOUR_API_KEY

Here is an example of how to use the library:

import asyncio
from limin_vision import process_image_from_url

async def main():
    url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/dd/Gfp-wisconsin-madison-the-nature-boardwalk.jpg/2560px-Gfp-wisconsin-madison-the-nature-boardwalk.jpg"
    result = await process_image_from_url(url)
    print(result)

if __name__ == "__main__":
    asyncio.run(main())

You can pass additional parameters to the process_image_from_url function to customize the model configuration and prompt:

await process_image_from_url(
    url,
    prompt="What's in this image?",
    model_configuration=ModelConfiguration(
        model="gpt-4o",
        temperature=1.0
    ),
    detail="high",
)

You can find the full example in examples/process_from_url.py.

Alternatively, you can process an image from a local file by calling process_image_from_file instead of process_image_from_url.

from limin_vision import process_image_from_file

async def main():
    result = await process_image_from_file("image.png")
    print(result)

if __name__ == "__main__":
    asyncio.run(main())

Just like with process_image_from_url, you can pass additional parameters to the process_image_from_file function to customize the model configuration and prompt:

await process_image_from_file(
    "image.png",
    prompt="What's in this image?",
    model_configuration=ModelConfiguration(
        model="gpt-4o",
        temperature=1.0
    ),
    detail="high",
)

You can find the full example in examples/process_from_file.py.

You can also get a structured response from the model by passing a response model to the process_image_from_url_structured or process_image_from_file_structured functions.

For example, here's how you can process an image from a URL and get a structured response:

import asyncio
from limin_vision import process_image_from_url_structured
from pydantic import BaseModel

class ImageResponse(BaseModel):
    title: str
    description: str

async def main():
    url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/dd/Gfp-wisconsin-madison-the-nature-boardwalk.jpg/2560px-Gfp-wisconsin-madison-the-nature-boardwalk.jpg"
    result = await process_image_from_url_structured(url, ImageResponse)
    print(result)

if __name__ == "__main__":
    asyncio.run(main())

You can find the full example in examples/process_from_url_structured.py.

You can also process an image from a local file by calling process_image_from_file_structured instead of process_image_from_file.

from limin_vision import process_image_from_file_structured

async def main():
    result = await process_image_from_file_structured("image.png", ImageResponse)
    print(result)

if __name__ == "__main__":
    asyncio.run(main())

You can find the full example in examples/process_from_file_structured.py.