About sample and shift sample

[ltc576935585]

Hello, thank you very much for your team's work. When I was studying your work, I visualized multiple versions of the sampling and found that the v2 version of the sampling you used did not match the polygon sampling in some places with the original label. The sampling of v4 version will not have this problem, it is based on the original 20 points sampling directly for shift sampling. May I ask if you have noticed this issue? Or what are your considerations?

https://blog.csdn.net/qq_41928168/article/details/144714210?sharetype=blogdetail&sharerId=144714210&sharerefer=PC&sharesource=qq_41928168&spm=1011.2480.3001.8118

[vigorous0503]

same questions.
I am curious about the results for these different sample method.

Thank you.