Hello, in Table 1 of your paper, you test some combinations of augmentations. I'm more concerned about the generalization of models, so I have some questions.
Have you tested CES + VFlip?In Table 1, it seems like just only VFlip augmentation can greatly increase generalization. However, this table only shows that CES + HSFlip will decrease the generalization performance of CES.
You set the asymmetric probability of color jitter to zero in your code. However, I think asymmetric color jitter will increase the generalization performance. Have you tested CES (asymmetric color jitter)?
Hello, in Table 1 of your paper, you test some combinations of augmentations. I'm more concerned about the generalization of models, so I have some questions.
I hope to receive your reply soon.