Open DCNSW opened 3 years ago
The code shows that the PointASNL classification network adopts the first 1024 points as input by default, I am really confused about why the first 1024 points of the aligned modelnet40 dataset are the best?
Thank you if anyone answers this. @yanx27
The code shows that the PointASNL classification network adopts the first 1024 points as input by default, I am really confused about why the first 1024 points of the aligned modelnet40 dataset are the best?
Thank you if anyone answers this. @yanx27