search

mhamilton723 / STEGO

Unsupervised Semantic Segmentation by Distilling Feature Correspondences
MIT License
711 stars 142 forks source link

Experiment with Different Number of Classes #96

Open rd1886 opened 1 month ago

rd1886 commented 1 month ago

Hi, I'm running STEGO on my custom data. Do I need to re-run the pre-compute KNN each time I change the number of classes to predict?

Thanks!