ImageSorter

Visual Sorting by Color: ImageSorter automatically arranges images based on their color similarity, grouping visually similar images close together on a 2D map or a 3D globe, which makes it easier to navigate and locate specific photos using visual memory.

Interactive Navigation: Users can explore their image collections through an interactive interface, allowing them to zoom in, rotate, and view images in detail on a flat map or a rotatable globe, enhancing the browsing experience for large photo libraries.

Efficient Image Management: The software allows users to copy, move, or delete images directly within the interface, streamlining the organization of extensive image collections without the need for external file management tools.

To begin using ImageSorter, the first step is to install the software on your computer. Once installed, open the application and select the folders containing the images you want to sort. The software allows you to navigate your image collection using either a 2D map or a 3D globe, which are generated based on the similarity of images. As you explore the map or globe, you can zoom in on specific images by clicking on their thumbnails. If you are looking for a particular image but are unsure of its location, you can use ImageSorter’s visual organization to help narrow down your search. The software groups similar images closely together, making it easier to find what you are looking for based on visual memory. You can also use the software to perform additional tasks such as copying, moving, or deleting images directly from the interface. This functionality is particularly useful for organizing large collections of photos. ImageSorter does not require a specific query image, making it a versatile tool for browsing and managing images.

ImageSorter operates using a sophisticated image analysis process, which goes beyond simply calculating the average color of each image. The software calculates various features such as color structures, gradients, and color histograms to provide a detailed comparison between images. These calculated features allow ImageSorter to evaluate the visual content of images more accurately, enabling precise sorting and grouping. The sorting on the 2D map is achieved through a custom variant of the Self-Organizing Map (SOM) algorithm, specifically designed to project images onto a 2D grid. This adaptation allows ImageSorter to arrange similar images closer together while positioning dissimilar images further apart. The custom SOM creates a visually coherent and easily navigable layout, organizing images by their content rather than relying solely on metadata like file names or dates. Additionally, ImageSorter caches thumbnails, enhancing the speed and efficiency of browsing after the initial load. This combination of advanced image analysis and the custom SOM algorithm ensures efficient navigation through large image collections, making it easier to locate specific images based on visual similarity.

The x64-bit version of ImageSorter can allocate more than 4GB of memory, allowing it to handle more images than the 32-bit version. Since the download archive does not include an installer (no setup.exe), the contents must be extracted to the desired directory. This Windows-only version of ImageSorter is also available on GitHub. Contributions through issues and pull requests are welcome to improve the application's stability and functionality.

GitHub Repository:
https://github.com/Visual-Computing/ImageSorter