meispi / Findupe

This is an ML-based project which separates duplicate images from a bunch of images.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Findupe

A tool that helps you separate similar looking images from a bunch of images. It has 2 modes similar (default) and strict, in similar mode if 2 images are 90% or more similar on the basis of structural similarity index then they are considered to be duplicates. In strict mode, if 2 images are exact copy of each other only then they are considered to be duplicates.

Installation

Make sure you have python (or python3) installed on your system and have it set as an environment variable. You can get it from here

$ git clone https://github.com/meispi/Findupe
$ cd Findupe
$ python setup.py install

How to use

optional arguments:
  -h, --help  show this help message and exit
  -p PATH     source directory path
  -s          strict (only exact copies will be detected)

Here -p flag is necessary followed by the full path of the directory which contains the images. The program will create 2 directories, inside the current working directory, Dupes and Original(so make sure you don't have directories/folders of same name). Dupes will contain all the duplicates of an image (if there are n similar looking images then it will contain n-1 of them). Original will contain unique images.

-s is an optional flag (no argument required) which will result in separating only the exact copies of images (if 2 images are similar looking but not exactly the same, this will treat them as 2 different images).

e.g. :

For separating similar looking images

$ python findupe.py -p /path/to/dir

image

Dupes

image

Original

image

For separating exact same images

$ python findupe.py -p /path/to/dir -s

image

Dupes

image

Original

image

About

This is an ML-based project which separates duplicate images from a bunch of images.

License:MIT License


Languages

Language:Python 100.0%