image_comparator

Design

Technical choices for components:

• Queue - Python multiprocessing module shared joinable queue
• Consumers - Python multiprocessing module to distribute asynchronous tasks
• Storage - File system with shared distributed lock from python multiprocessing module
• Algorithm - ImageHash

Installation

Docker (Windows, Linux, MacOS)

• run docker build --tag image_comparator:1.0 . to build the image
• run docker run -it image_comparator:1.0 to start the container. It will provide a command line shell inside the container
• run docker run -it image_comparator:1.0 to start the container
• run python image_comparator.py images.csv to run image comparison on sample images provided in 'images' folder

Virtualenv (Linux, MacOS)

• run virtualenv venv to create a virtual environment
• run source venv/bin/activate to start the virtual environment
• run pip install -r requirements.txt to install all requirements
• run python image_comparator.py images.csv to run image comparison on sample images (provided in 'images' folder)

Output

The results can be found inside 'results' folder.

• The name of the output file is in the format output-.csv (for eg: output-1589203893333.csv)
• It is a csv file with 4 columns "image_1", "image_2", "similar", "elapsed" in the same order

How to use this for your own images

Copy your images to the 'images' folder. Edit 'images.csv', leave the header and edit the rows to add images to compare. A few things to consider:

• In 'images.csv', use 'images/' prefix with all image paths. It is relative path since images are inside 'images' folder.
• Absolute paths would work too. But these would be different when running on your host machine and inside the docker container (working directory inside docker container is '/usr/src/app').
• You can use a different folder to save images and a different name for the input csv file. The csv file should have the 2 headers (image1, image2). If using docker container, you will need to update the 'Dockerfile' to copy the new folder and new csv file into the container.
• If using docker, everytime a change is made, you will need to rebuild the docker image before running (docker build --tag image_comparator:1.0 .)
• Syntax to run the program is python image_comparator.py <image_csv_file>