Guided clustering UNI model subpatch embeddings. With a human in the loop, you can define semantically meaningful cluster centers, based on the features extracted with a UNI model. The cluster centers are then used as a segmentation model for other slides. Feature extracting model can be swapped for something else in src/model.py.
To run the project you should run cluster_subpatch.ipynb. It uses samples from data/med_examples, but you can add your own .tiff files.
Currently only .tiff files are supported, but other file extensions are WIP.