Open jpsml opened 2 years ago
Hello, I am also having some problems reproducing Campus dataset results, the pre-trained weight did not work either. I am getting a very similar accuracy: average 90.8%. Did you figure out how to modify the training parameters to get it to work on the Campus dataset?
No, I was not able to reproduce the paper results.
Hi, I have changed the mvp_campus.yaml file for setting END_EPOCH to 40 in order to try to reproduce the results reported in the paper. However, I got the results below:
PCP | 91.02 | 93.02 | 88.48 | 90.84 | Recall@500mm: 1.0000
Should I change anything else in order to reproduce the paper results for the Campus dataset? Thanks.