Request of evaluation sample indices and mesh results

[Sangminhong]

Thank you for your great paper!
As I delved into your research, I came across the section where you mentioned the selection of 50 shapes from ABC, with 20 from each category of ShapeNet for evaluation.

I am eager to learn more about the specific samples you have chosen for evaluation, and I was wondering if you could kindly provide me with the index numbers or any additional details regarding these selected shapes.

Furthermore, I am keen to gain a better understanding of your work and would greatly appreciate it if you could share the mesh qualitative results as presented in your paper.

[FENGGENYU]

Hi, Please check our latest version on CVPR website.

https://openaccess.thecvf.com/content/CVPR2022/html/Yu_CAPRI-Net_Learning_Compact_CAD_Shapes_With_Adaptive_Primitive_Assembly_CVPR_2022_paper.html

We used 1000 shapes from ABC and 1300 shapes from ShapeNet.

The processed data are given in the data link on the readme.

[FENGGENYU]

Here are three examples I just tried after running my codes:

https://drive.google.com/drive/folders/1SZZLqQ6maMLpDVxIzWtEUNad0SqWrW1V?usp=share_link

You can try my codes to get more results.

[Sangminhong]

Thank you for your kind reply. Could you also provide the evaluation metrics (CD, ECD, NC, LFD) implementations?

[FENGGENYU]

You can find them on the BSP-Net evaluation section:
https://github.com/czq142857/BSP-NET-original