search

serizba / salad

Optimal Transport Aggregation for Visual Place Recognition
GNU General Public License v3.0
140 stars 13 forks source link

DINOV2-Netvlad #4

Closed pQWQq closed 2 weeks ago

pQWQq commented 6 months ago

Hello, I want to reproduce the experiment in the paper, but cannot find the corresponding parameters. Can you provide the training code of DINOV2-Netvlad?

serizba commented 3 months ago

Hi @pQWQq , you can use the same hyperparameters for training. Just append a NetVLAD aggregator.