"dbscan": {"file_extension": ".txt", "folder_name": "dbscan", "nb_features": 699 / 2}, # VideoMem
I am getting only 540 nb_features including both .
And the spearman_correlation of 0.46.
Also in your code in gen_features.py
"hrnet" is hardcoded , are we not calculating similarity for both hrnet and ipcsn152 through dbscan
In my case similarity features are not contributing in anything. Do you have any suggestions regarding this