Some questions about the data augmentation step

[Vote2Cap-DETR]

[Vote2Cap-DETR]

In my experiments, the data augmentation step seems to have a negative effect on the accuracy in the ModelNet40 experiment. Have you encountered the same problem?

[MenghaoGuo]

Hi, I do not have the same question. Moreover, I want to know which data augmentation makes performance drop in your experiments.

[Vote2Cap-DETR]

I performs the same augmentation as you mentioned in your paper: [-0.2, 0.2] translation, [-0.67, 1.5] scaling, and random input dropout. (does the random input dropout mean a uniform sampler of input points?)