search

zkcys001 / UDAStrongBaseline

Open-source stronger baseline for unsupervised or domain adaptive object re-ID.
142 stars 21 forks source link

A question when I run the code of GLT Stage II #28

Open pumpkin-vera opened 2 years ago

pumpkin-vera commented 2 years ago

==> Create pseudo labels for unlabeled target domain Traceback (most recent call last): File "sbs_trainkmeans.py", line 537, in main() File "sbs_trainkmeans.py", line 166, in main main_worker(args) File "sbs_trainkmeans.py", line 296, in main_worker cluster.train(moving_avg_features) File "/home/lzshang/anaconda3/envs/slz_udaabs3/lib/python3.7/site-packages/faiss/init.py", line 734, in train clus.train(x, self.index, weights) File "/home/lzshang/anaconda3/envs/slz_udaabs3/lib/python3.7/site-packages/faiss/init.py", line 85, in replacement_train self.train_c(n, swig_ptr(x), index) File "/home/lzshang/anaconda3/envs/slz_udaabs3/lib/python3.7/site-packages/faiss/swigfaiss.py", line 5059, in swig_ptr return _swigfaiss.swig_ptr(a) ValueError: input not a numpy array

I have check the type of moving_avg_features, is numpy.ndarray