Thank you for sharing your excellent research work and results.
I have a question regarding the training configuration.
In the README, under the Training section, the copy_paste probability is specified, which indicates that this augmentation is used.
However, in the coco.yaml file used for training, the setting segments=False seems to disable the use of copy_paste.
Could you clarify which setting is correct?
Additionally, were the models provided in the Validation section trained with segments=True?
Thanks in advance for your help and clarification!