https://github.com/itsluketwist/imageai-wizard
ImageAI Wizard is a Python library that provides various image utilities using the OpenAI API.
https://github.com/itsluketwist/imageai-wizard
image-processing openai-api
Last synced: 3 months ago
JSON representation
ImageAI Wizard is a Python library that provides various image utilities using the OpenAI API.
- Host: GitHub
- URL: https://github.com/itsluketwist/imageai-wizard
- Owner: itsluketwist
- License: mit
- Created: 2023-03-28T18:23:57.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2024-01-13T08:19:36.000Z (over 1 year ago)
- Last Synced: 2025-01-24T22:29:23.748Z (5 months ago)
- Topics: image-processing, openai-api
- Language: Python
- Homepage:
- Size: 21.5 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# **imageai-wizard**
ImageAI Wizard is a Python library that provides an easy-to-use interface for generating
various data about images using the OpenAI API.
## *installation*
ImageAI Wizard can be installed via pip:
```shell
pip install git+https://github.com/itsluketwist/imageai-wizard
```
## *usage*
First, you need to obtain an API key from OpenAI (use [this](https://platform.openai.com/account/api-keys)
interface, you must be [signed up](https://beta.openai.com/signup/)).
This key must be provided when initialising the `Wizard` class, or stored in the
`IMAGEAI_WIZARD_OPEN_AI_KEY` environment variable. The code will error if no key is provided.
```python
from imageai_wizard import Wizard, Persona, Tone
wiz = Wizard(api_key='YOUR_API_KEY')
image_url = "https://upload.wikimedia.org/wikipedia/en/thumb/4/47/Iron_Man_%28circa_2018%29.png/220px-Iron_Man_%28circa_2018%29.png"
# generate text based responses about the image
description = wiz.generate_description(
image=image_url,
persona=Persona.CHILD,
tone=Tone.CHEERFUL,
)
print(f"Description: {description}")
caption = wiz.generate_caption(
image=image_url,
persona=Persona.ACADEMIC_PROFESSOR,
tone=Tone.ANGRY,
)
print(f"Caption: {caption}")
title = wiz.generate_title(
image=image_url,
persona=Persona.BUSINESS_PERSON,
tone=Tone.CONFUSED,
)
print(f"Title: {title}")
# generate a comparison to another image
another_image_url = "https://e0.pxfuel.com/wallpapers/971/526/desktop-wallpaper-robot-and-cat-red-robot.jpg"
comparison = wiz.compare_images(
image_1=image_url,
image_2=another_image_url,
)
print(f"Comparison: {comparison}")
```
## *development*
Clone the repository code:
```shell
git clone https://github.com/itsluketwist/imageai-wizard.git
```
Once cloned, install the package locally in a virtual environment:
```shell
python -m venv venv
. venv/bin/activate
pip install -e ".[dev]"
```
Install and use pre-commit to ensure code is in a good state:
```shell
pre-commit install
pre-commit run --all-files
```
## *contributing*
If you'd like to contribute to ImageAI Wizard, please fork the repository and create a pull request.
We welcome contributions of all kinds, including bug fixes, new features, and documentation improvements.
## *license*
ImageAI Wizard is licensed under the MIT License. See [LICENSE](LICENSE) for more information.
## *inspiration*
I wanted to get more familiar with ChatGPT, it's API and writing prompts in general.
So I did the only reasonable thing, and asked ChatGPT for a python library I could
build that utilises the API! This was the suggestion, ChatGPT even helped me to
write the code itself, and this readme.
Initial prompt: `Can you suggest a simple python library that I can build? the library must use your own API and be useful.`
## *todo*
- Ask ChatGPT for more ideas... and implement them, I guess?
- Add some image generation methods that utilise `openai.Image.create(...)`