How to handle the repeated characters in the inference of trained recognition models? #6022
JimEverest
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
You can check whether the dimension of training is the same as that of inference. If the dimension of training is [3, 32, 320], and the dimension of inference is [3, 64, 640], there are many repeated characters be ng recognized.
Beta Was this translation helpful? Give feedback.
All reactions