missing .npz files for famous dataset

[Linusnie]

hi, many thanks for you work on this method

I'm looking into reproducing your results on the Famous dataset. But I noticed that the .npz files are missing from the download link in the readme (https://drive.google.com/drive/folders/1qre9mgJNCKiX11HnZO10qMZMmPv_gnh3?usp=sharing). Would you be able to add those files in?

I also tried recreating the point clouds with python sample_query_point.py --out_dir /home/linus/workspace/data/neural_pull/famous_new/ --input_dir /home/linus/workspace/data/points2surf/famous_noisefree/04_pts/ --dataset famous, but I get the error

/home/linus/workspace/data/neural_pull/famous_new/3DBenchy.npz
Traceback (most recent call last):
  File "/home/linus/workspace/NeuralPull/sample_query_point.py", line 124, in <module>
    point_idx = np.random.choice(pointcloud.shape[0], POINT_NUM_GT, replace = False)
  File "numpy/random/mtrand.pyx", line 1000, in numpy.random.mtrand.RandomState.choice
ValueError: Cannot take a larger sample than population when 'replace=False'

Since the 3DBenchy dataset has less than POINT_NUM_GT(=20000) points. Could you clarify what value of POINT_NUM_GT was used for the famous dataset?

[kbrian8]

HI @Linusnie , did you manage to resolve the above issue? I'm facing the same.

[Linusnie]

@kbmufti kind of, I moved on to using the pytorch version, where the pre-processing is applied automatically if you pass in a .xyz file see here

I should say the sampling settings in the pytorch repo are slightly different from the tensorflow implementation, so it would still be nice to have the original .npy files for reproducing the paper results. For example: in the pytorch repo the local sigma scale is set automatically based on the number of points, while it's fixed in tensorflow. The pytorch repo also auto-scales the point cloud to fit in a 1x1x1 bounding box centered at 0.

[kbrian8]

@Linusnie , Thank you for you response, What version of python are you using? I am getting dependencies conflicts

[Linusnie]

I followed the instructions here #2 (comment)