How many points are used for shape classification on the ModelNet40 in training phase?

Question

DapengFeng opened this issue 3 years ago · 4 comments

Thanks for your excellent work. I get a question about the number of points used for shape classification on the ModelNet40 after reading the paper.

Answer 1 · 2021-12-15T12:26:44.000Z

usually, 1024 for classification, 2048 for segmentation. This is my experience.

Answer 2 · 2021-12-15T14:23:53.000Z

Thanks for your reply.

Answer 3 · 2021-12-18T01:22:36.000Z

Can you reproduce the results reported by paper?

Answer 4 · 2021-12-18T06:41:29.000Z

Can you reproduce the results reported by paper?

No, I have not done it. I just read some papers related to point transformer recently