YOLOX: Accuracy Issues When Retraining YOLOX on the YCBV Dataset

[piyanuoyubuliao]

Thank you for your outstanding work. I tried retraining the YOLOX model on the YCBV dataset using the default parameters, but I couldn't achieve the expected accuracy. Could it be due to changes in hyperparameters or other details? Thanks!

[piyanuoyubuliao]

I am currently working with the YOLOX-X model and have trained it using the default parameters. However, I was unable to reproduce the ADD-S performance results reported in the original paper.
Attached is the ADD-S curve from my training for your reference. I would be very grateful if you could kindly share more detailed information about your training procedure, as well as the train_log.txt file if available.
Thank you very much for your time and support.