evaluate with 256 points

[azhua66]

hello! I get a model with pointnet_cls_ssg.py,When I evaluate with 256 input points,I get 58% accuracy ,it is not the same with the paper which is much higer than 85%?why?

[lyy-1997-hub]

你好！我得到了一个具有pointnet_cls_ssg.py的模型，当我用256个输入点进行评估时，我得到了58％的准确度，这与85％以上的论文不一样？

why don't use 1024?

[azhua66]

你好！我得到了一个具有pointnet_cls_ssg.py的模型，当我用256个输入点进行评估时，我得到了58％的准确度，这与85％以上的论文不一样？

why don't use 1024?

result evaluate with 1024 is almost as same as the paper mentioned. I use 256 points just want to test the robustness of the model , in this paper when evaluate with 256 it can get a high accuracy but i got 58% accuracy.