Skip to main content

A tool to compare similar images

Project description

imgsimilarity

Python package

imgsimilarity tool automates the process of comparing image pairs to check if they are similar or not and produces a score that quantifies the comparison by providing a score between 0 to 1, 0 being exactly identical.

How to use the tool

The tool is published in the pypi repository and the latest version can be installed with just a single line of command.

Prerequisites

  1. Python should be installed version 3 and above.
  2. Python package manager "pip" should be installed.

Installation

Open a terminal and paste:

pip install imgsimilarity

This will download the tool and set it up by installing all its dependencies

Execution

The module imgsimilarity takes a command line argument, i.e the path to the location of the csv file which needs to be provided when running it like in the example below

python -m imgsimilarity.diff C://pathtocsvfile

The tool will execute and a comparison_results.csv file will be generated in the directory the command was run.

How does the tool work?

The tool uses dhash algorithm to find a hashcode for each image and then find how many bits differ between the hashcodes to find the score for similarity. Identical images have difference as 0.

The dhash pypi module is being used here, its an out of the box python module to compute hashes and determine the difference for images. The hash for an image is stored in a dictionary to avoid repeated execution of dhash function and this reduces the cost of similarity detection. The execution time for each pair is calculated by taking the difference of process time at the start of the iteration and the process time at the end of the iteration.

Below is a flowchart that briefly illustrates how the tool functions.

alt text

How does the tool achieve hassle free installation and usage?

It is by using the Python Package Index repository!

Once the latest code is pushed and tagged, the gitlab ci is configured to run pytests to make sure the code works and the once the tests succeed it creates the build and a distributable with the help of setup.py file and twindle. and its uploaded to the PyPi Repository automatically.

The user can install the latest version using pip install or upgrade command. Thanks to the setup.py file all the dependencies are installed automatically

alt text

Reference:

PyPi repo: https://pypi.org/project/imgsimilarity/ dhash : https://pypi.org/project/dhash/

note

The source code is well documented and pydoc can be genrated.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

imgsimilarity-1.tar.gz (4.4 kB view hashes)

Uploaded Source

Built Distribution

imgsimilarity-1-py3-none-any.whl (4.5 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page