You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi,
Thanks for your working. I am working with ur baseline with the pretrained model u provided and have a issue to find the threshold to determine whether the recognized person is right or not. I added re-ranking after getting the q-g-dist matrix and set the threshold as 0.1. Is there any recommand value for this threshold?
Thx!
The text was updated successfully, but these errors were encountered:
Thanks for your attention. I have not conducted recognition, just using it for retrieval experiments. The threshold for recognition should be based on intra-class and inter-class distances. You may have to compute some statistics on the whole val or test set to fulfill your purpose.
Hi,
Thanks for your working. I am working with ur baseline with the pretrained model u provided and have a issue to find the threshold to determine whether the recognized person is right or not. I added re-ranking after getting the q-g-dist matrix and set the threshold as 0.1. Is there any recommand value for this threshold?
Thx!
The text was updated successfully, but these errors were encountered: