search

Yochengliu / Relation-Shape-CNN

Relation-Shape Convolutional Neural Network for Point Cloud Analysis (CVPR 2019 Oral & Best paper finalist)
https://yochengliu.github.io/Relation-Shape-CNN/
MIT License
415 stars 71 forks source link

Could you release your trained model? I just run your code, but I just get 84% cls accuracy. #10

Closed liruihui closed 5 years ago

ShiQiu0419 commented 5 years ago

No description provided. I have got 92.2% with the default settings. (P.S. your work on PU-GAN is excellent, thanks for sharing)

liruihui commented 5 years ago

No description provided. I have got 92.2% with the default settings. (P.S. your work on PU-GAN is excellent, thanks for sharing)

Thanks. But the report accuracy is 93.6%, which makes a big improvement, compared with other methods. 92.2% is also not enough.

lqzhao commented 5 years ago

@liruihui image Without voting and random scaling tricks, it achieves 92.2% acc.

liruihui commented 5 years ago

@liruihui image Without voting and random scaling tricks, it achieves 92.2% acc.

Thanks.

Ailce8862 commented 4 years ago

No description provided. I have got 92.2% with the default settings. (P.S. your work on PU-GAN is excellent, thanks for sharing)

Thanks. But the report accuracy is 93.6%, which makes a big improvement, compared with other methods. 92.2% is also not enough.

Did you finally get the accuracy of the paper

Ailce8862 commented 4 years ago

@liruihui image Without voting and random scaling tricks, it achieves 92.2% acc.

How do I get the accuracy of the paper

Ailce8862 commented 4 years ago

No description provided. I have got 92.2% with the default settings. (P.S. your work on PU-GAN is excellent, thanks for sharing)

Did you finally get the accuracy of the paper

Ailce8862 commented 4 years ago

@liruihui image Without voting and random scaling tricks, it achieves 92.2% acc.

Thanks.

Did you finally get the accuracy of the paper

liruihui commented 4 years ago

around 92%, but not stable.

Ruihui Li

Ailce8862 commented 4 years ago

thanks

梁攀

邮箱：liangxiaopan8862@126.com |

签名由 网易邮箱大师 定制

On 01/04/2020 11:10, Li Ruihui wrote:

around 92%, but not stable.

Ruihui Li

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

sheshap commented 4 years ago

For SSG I got 92.71 as mentioned in the GitHub with voting. In the paper, the same has been mentioned as 92.2.

Anybody tried for msg please let us know the radii which you used.