The purpose of the duplicate searcher is to search your image collection for duplicate images. It has many features to help with making removing duplicate images from your collection easier, including:

• Recursively searching folders in your selected directory.
• Toggles for searching images or videos or both at the same time.
• Multi-threading to speed up searching with customizable thread count.
• Similarity threshold will only show potential duplicates above that percentage.
• File search scope and matches count display.
• Real time searching allows to sort through found duplicates while the program is still searching.
• File previews in the grid view allows for quicker searching.
• File info
  • Display of dimensions for images and videos.
  • Display of video duration.
  • Comparing file "tags" (words in the filename which could be used as tags)
    • A tag in bold indicates it is not on the other file.
    • A tag in italics indicates that the tag is in your program's list of custom made tags by you.
    • A tag in both bold and italics indicates that both of the above are true.
• Support for merging file tags if choosing one image (left or right) is selected
  • In the image below, the resulting filename would be: "by cat dog dogs eaten pulls skin.jpg". Not particularly useful but if you are sorting images that have been tagged using the tag convention encouraged by the sorter function then this is a very useful feature to have.

As seen below in the list of potential duplicates, the first and second row look to be duplicates, but the 3rd and 4th rows do not, but they still have a high similarity chance, this is because the algorithm used is very simple and sometimes doesn't work very well with certain images or collections of images, this is why the program is only used to show you potential duplicates, and doesn't take deleting them into it's own hands, as it would risk deleting many false positives, losing data.

The choice of algorithm used is a feature that that I would like to implement in the future, but currently not in the program.