I built a small, self-contained K-Means implementation in pure PyTorch: https://gitlab.com/hassonofer/pt_kmeansI was working on dataset sampling and approximate nearest neighbor search, and tried several existing libraries for large-scale K-Means. I couldn't find something that was fast, simple, and would run comfortably on my own workstation without hitting memory limits. Maybe I missed an...
If you're interested in trying a web-based alternative, I have a demo space at https://huggingface.co/spaces/birder-project/birder-image-cl... where you can upload images directly from your browser. Fair warning though - it's primarily designed to showcase the models rather than provide any kind of user experience, so you'll need to manually select the appropriate regional model (look for suffixes like eu-common, arabian-peninsula, etc. based on where your photos were taken).
The coverage is still very limited... More regions will follow in the coming months.