This is a parallelized implementation of the mean shift clustering algorithm through TensorFlow. This makes it easy to use on the GPU if you install TensorFlow with CUDA support.
I wrote this to better understand tf.while_loop and how to implement a nontrivial iterative algorithm in a TensorFlow graph. It only works in TensorFlow v1, or through the tensorflow.compat.v1 module of TF2.