Skip to content

A digital album face recognition manager, that isolates images of a specified person from a digital album.

License

Notifications You must be signed in to change notification settings

grebtsew/AlbumOrganizer

Repository files navigation

AlbumOrganizer

AlbumOrganizer is a Python3 tool that organizes your photo albums by recognizing persons in your photos and creating a file structure where photos of each individual are moved. The implementation is dockerized and uses pytest for testing. Extra features include generation of face collages per person and sentient slideshows, see more below. The implementation utilizes multiprocessing and checkpointing in order to optimize and maintain long running tasks due to large photo albums. The demo below visualizes the execution of the main function on test images.

license size commit Version Downloads

Demo

Installation

The implementation can be executed either using Docker of locally on your machine, the two install processes are described below.

Docker

To install AlbumOrganizer using Docker, you need to have Docker installed on your system. You can then build and run the container using the docker-compose file using the following command:

docker-compose build albumorganizer

docker-compose run albumorganizer <command (optional)>

NOTE: The optional is used to run separate feature functions and tests.

Locally

To install AlbumOrganizer locally, follow the steps in the DockerFile. Python3.8 is recommended as it was used during development. Run the following command to install python packages:

pip3 install -r ./requirements.txt

Usage

Replace album_path with the path to your photo album. AlbumOrganizer will scan your photos, recognize persons, and create a file structure where photos of each individual are moved to a separate folder. The general execution pipeline includes the following steps:

  1. Copy photo album to the ./data folder for docker volume. (Optional).
  2. Update album root path and settings in ./main.py.
  3. Run the implementation by executing ./main.py.
  4. All results will be stored in ./target folder.

Using docker the implementation, simply execute the following command in your terminal:

docker-compose run albumorganizer

Locally execute the ./main.py file by running the following command:

python3 ./main.py

Additional features can be found in the ./features folder.

Extra Features

These features can be found in the ./features folder. Change settings the files before running.

Face Collages

AlbumOrganizer can also generate face collages per person. This function will crop all faces of a specified person in a photo album and merge them together resulting in a large collage of faces. To generate face collages, simply add the --collage flag to the command:

docker-compose run albumorganizer --collage

AlbumOrganizer will then generate a face collage for each individual in the photo album.

Sentient Slideshows

AlbumOrganizer can also create sentient slideshows. A sentient slideshow is a slideshow that has been generated using a multitude of filters depending on Computer Vision. See the script file in ./features to see all filters. To create a sentient slideshow, simply add the --slideshow flag to the command:

docker-compose run albumorganizer --slideshow

AlbumOrganizer will then create a sentient slideshow using the photos in the photo album.

Resize All Images

AlbumOrganizer can also resize all images in an album, either crop, resize or rescale. To resize the dataset, simply add the --imageresize flag to the command:

docker-compose run albumorganizer --image-resize

AlbumOrganizer will then create a new folder in target containing the resized images.

Remove Duplicate Images

AlbumOrganizer can also detect and remove duplicate images in albums. To remove duplicates, simply add the --removeduplicates flag to the command:

# Detect only (optional)
docker-compose run albumorganizer --detect-duplicates
# Remove
docker-compose run albumorganizer --remove-duplicates

AlbumOrganizer will then create a sentient slideshow using the photos in the photo album.

Testing

AlbumOrganizer comes with a set of tests that can be run using pytest. To run the tests, execute the following command in your terminal:

docker-compose run albumorganizer --pytest

Contributing

If you would like to contribute to AlbumOrganizer, please fork the repository and create a pull request. We welcome all contributions, big or small!

Disclamer

Some of the code in this implementation is generated using ChatGPT4.

License

AlbumOrganizer is released under the MIT License.

Copyright (c) 2023 Grebtsew