Open
Description
After training a custom model and converting it to NCNN, I'm getting unexpected weird detections as below. There were no issues with torch or ONNX inference. Anyone able to solve it?
When I convert the given pretrained model to NCNN, it works fine. This is strange behaviour. I suspect it has something to do with code changes that happened after 2021
Metadata
Metadata
Assignees
Labels
No labels
Activity