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
I am curious why test score thresholds are set so low in most config files (about 0.05). My understanding is that during bbox evaluation this threshold is varied anyways to obtain the precision-recall curve, then calculate the area under the curve and obtain the average precision (AP). However, this should lead to a high false positive rate during inference.
Is the score threshold set so low just to get higher AP scores during testing?
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Hello everyone,
I am curious why test score thresholds are set so low in most config files (about 0.05). My understanding is that during bbox evaluation this threshold is varied anyways to obtain the precision-recall curve, then calculate the area under the curve and obtain the average precision (AP). However, this should lead to a high false positive rate during inference.
Is the score threshold set so low just to get higher AP scores during testing?
Thanks
Beta Was this translation helpful? Give feedback.
All reactions