FaceCropThumb is a Python package for generating thumbnails of detected faces in images. It utilizes the MTCNN (Multi-Task Cascaded Convolutional Neural Network) for accurate face detection and OpenCV for image processing.
All documentation is in the “docs” directory and online at https://facecrop-thumb.readthedocs.io/en/stable/. If you’re just getting started, here’s how we recommend you read the docs:
- Detect faces in images using the MTCNN model
- Crop detected faces with a configurable margin
- Generate square thumbnails of detected faces
- Option to resize the entire image if no face is detected or face detection is skipped
- Command-line utility for easy usage
You can install FaceCropThumb via pip:
pip install facecropthumb
facecropthumb <image_path> [-d/--dir <output_directory>] [-m/--margin <margin_size>] [-F/--no-face] [-S/--skip-face] [-a/--all]
<image_path>: Path to the input image file.-d/--dir <output_directory>: Directory to store the output thumbnail. Default is the current directory.-m/--margin <margin_size>: Margin around the detected face. Default is 50 pixels.-F/--no-face: Skip face detection and resize the whole image if no face is detected.-S/--skip-face: Skip face detection and resize the whole image.-a/--all: Process all images in the specified directory. If no directory is provided, it processes images in the current directory.-v/--version: Show installed version.-h/--help: Show help.
Example:
facecropthumb input_image.jpg -d output_directory -m 30 -F
This command will generate a thumbnail of the detected face in input_image.jpg, with a margin of 30 pixels, and if no face is detected, it will resize the whole image to a square of 74 x Y pixels (~approx).
DISCLAIMER: This documentation is for the contributers of FaceCropThumb, not for general package users.
Firstly install dev dependencies from requirements.txt file
To run and test the package, use python -m facecrop-thumb <args>
You can see the production build docs in docs/build/index.html
Update the documentation at any cost after any change in feature
Run this to watch the docs/ folder while you're making documentation changes
watchmedo shell-command --patterns="*.rst;conf.py;_templates/*" --recursive --command='sphinx-build -b html docs docs/_build' docs
After a successful change in the package, it's time to publish it officially.
-
Update the version in
facecrop_thumb/version.py -
Create a new release on release page. (This will automatically trigger GitHub actions to release a new version on PYPI)
That's it, the newer version will be available to everyone.
But if you want to publish a version by yourself, you can use following command:
-
python setup.py sdist -
twine upload dist/*
This project is licensed under the MIT License - see the LICENSE file for details.