You may have a lot of duplicate images on your hard drives. Now Exact duplicate images are easy to find and delete with
fdupes, but what about images that are different size, cropped, or different just a bit visually ?findimagedupes is a command line tool that finds visually similar images, even if they are of different resolution and size. There is a threshold setting to define how much similar two images have to be.Install findimagedupes with
apt-get install findimagedupesThe simplest form to run is
findimagedupes -R ~/imagesthis finds all the duplicate images recursively in all subdirectorie of
~/images.A more complete form which I generally use is
findimagedupes -v=fp -R -f=fp_data ~/images</div>
- -v : verbose mode. You can see fingerprint of eachfile with fp, or md5 hash with md5. For some reason the option -v=LIST specified in the command help doesn’t work. You only can get -v=md5 or -v=fp to work.
- -R : recursively search all subdirectories
- -f : This is a file to we are going to write to / or read from , each images fingerprint . especially useful for large directories with lots of images.
Generally I would let it run first for sometime so that I have all the files scanned and their signatures saved in
fp_data. On next run, the program reads off the fp_data file and runs a lot more faster.To use a photo viewer like feh, use the -p switch with full path . This is what you should do if you want to see the pairs and delete one or more of them. Feh lets you navigate in each set , delete the images and then move on to the next set.
-p doesn’t work with simple program name, even if this exist in your system path. You have to always specify the full path of the program.
findimagedupes -v=fp -R -f=fp_data -p= /usr/bin/feh ~/imagesYou can also set the threshold. To find Exactly same visual image, set the threshold to 100 for example
findimagedupes -v=fp -R -f=fp_data -p= /usr/bin/feh -t=100 ~/images