Training on my own data

[zyh10355]

I would like to ask, if you train on your own data, are there any sampling requirements or things to pay attention to? If applied on the part, then for sampling also directly scan the sample as input? and How many samples are needed in total as a training set for a kind of part? Thank you!!!

[zyh10355]

Thank you very much for your code. I used your code to reproduce the point cloud feature extraction on my own data set. The number of point cloud samples is 2280, and the verification set is 120.