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
when I use zero-shot pre-trained model to run my own videos, the model output 10 instances which is highly redundant(usually highly overlapped at one object),how can we choose the best object result?
The text was updated successfully, but these errors were encountered:
In that case, the reason could be that the model is not able to find any other relevant instance in the image and I would choose the first mask given by the decoder.
when I use zero-shot pre-trained model to run my own videos, the model output 10 instances which is highly redundant(usually highly overlapped at one object),how can we choose the best object result?
The text was updated successfully, but these errors were encountered: