K means clustering for images

Usage

  cargo run <path to image> <k>

Example:

Input Image	Output (k = 50)

Limitations

• Arbitrary n-dimensional data isn't supported and isn't planned to be supported. This is only really meant for image segmentation
• No convergence tracking yet, currently the number of k-means iterations is hardcoded
• K (#clusters) must be specified manually at the moment

Planned

• Add support for reading-out results from the GPU. Currently the results can only be viewed visually
• Auto-select k instead of specifying it manually, using the Elbow Method or similar
• Convergence tracking