Closed naoto0804 closed 4 months ago
Given n text features, RADM may generate a non-n number of elements for text. Is there any way to post-process or filter the outputs? I'm training on RADM on my dataset and facing the aforementioned issue.
n
You can first use an NMS operation, then sort the results in descending order of confidence, and select the topN box according to your needs.
Given
n
text features, RADM may generate a non-n
number of elements for text. Is there any way to post-process or filter the outputs? I'm training on RADM on my dataset and facing the aforementioned issue.