Reproduce the effect

[PjBhLy]

Hello, I have reproduced why the effect of the paper is particularly poor. There is no paper conclusion on the effectiveness of training speed, prediction rate, recall rate, mAP, etc. Do you know the general situation?

The former is the paper model, while the latter is the original yolovn8.pt

[mkang315]

I am not sure what happened to you. I suppose there may be something wrong with your configuration settings. Our model is based on YOLO8x (version 8.0.190).