Hi, @SilenKZYoung
Thanks for the great work. I am interested to try it and explorer more on this paper, but the link of Dataset is not shareable, which needs access and I have requested(yj444757397@gmail).

Thanks & Regards,
Jie

Thanks for the wonderful work! Could you share your code about the segmentation? How to align shapenet part labels to the segmentation results?

Hi, Kai Zhi, great job. And I have a question about compactness loss after reading the paper. M is a hyperparameter and fixed, why should we use a loss function to reduce the number of parts?
Thanks again, looking forward to your reply.

Hi @SilenKZYoung

I'm working on the code and trying to reconstruct your results. The pretrained checkpoints work fine and create the described results in your paper.

But I discovered, that the W_CST-Loss ist set to 0 per default in training.

parser.add_argument ('--W_CST', default = 0.00, type = float, help = 'CST loss weight, this loss is only for generation application')

Does it hurt the performance of the model? I could also not find any note of it in the published paper.

Thus the cuboid vector $p_m$ is only trained for rotation $r_m$, scale $s_m$ and existence $\delta_m$, and the translation $t_m$ does not appear in the loss.
This leads to the position of the cuboids to be randomly distributed in the 3D space, which is not the case with the pretrained checkpoints.

This design decision seems somewhat counterintuitive. Could you provide some ratio on this choice?

Hi -- awesome work!

I was just wondering if you were planning on releasing pre-trained models for any categories from ShapeNet? If you are, do you have any idea on when you might do so?

Thanks!

As stated in the title，Can change the code to extract the plane?

Hello, I encountered a little problem when trying to reproduce the reconstruction performance indicators in your paper. Since there is no such part in the project code, I ask a question here and hope to get your reply. Thank you very much!
Based on the pre-trained model, I uniformly sampled the generated geometry according to the method of the paper to obtain a predicted point cloud of 4096 points, and calculate the Chamfer Distance between it and the original point cloud (I used the pcu library to Calculation, link .)
The result obtained is roughly an order of magnitude away from yours (0.04 (in the paper) vs 0.004 (my experiment result)).
I Note that the point cloud is scale-transformed，as mentioned in the paper, "All shapes are pre-aligned and normalized to the unit scale", I am not sure if there are some other operations before calculating the CD distance.
What should I do to get the CD index in the paper? Thanks very much!

Attachment: The left side of the figure is my result after uniform sampling, and the right side is the original point cloud result. I calculated the CD between these two point clouds.

Hi, your results look very good, how did you render the final visual effect?

Hi @SilenKZYoung ,

I currently tried to evaluate your model. Unfortunately, the batch size of 32 does definitely not fit into 11GB of VRAM, not even 16. I could run the training only on a batch size of 8.

I used a RTX 2080Ti. Could you please tell me, how did you fit your Model on a GTX 1080Ti ?

I will try to get my hands on an GTX 1080Ti and try it once again.

ERROR MESSAGE:

Traceback (most recent call last):
  File "/clusterstorage/gkobsik/learning-relationships/./CuboidAbstractionViaSeg/E_train.py", line 344, in <module>
    main(args)
  File "/clusterstorage/gkobsik/learning-relationships/./CuboidAbstractionViaSeg/E_train.py", line 123, in main
    loss, loss_dict = loss_func(points, normals, outdict, None, hypara)
  File "/clusterstorage/gkobsik/anaconda3/envs/learning_relationships/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1190, in _call_impl
    return forward_call(*input, **kwargs)
  File "/clusterstorage/gkobsik/learning-relationships/CuboidAbstractionViaSeg/losses.py", line 62, in forward
    randn_dis = (torch.randn((batch_size,num_points)) * self.std).cuda().detach()
RuntimeError: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

作者您好