Kaolin installation problems with google colab

[feboz]

I'm following the steps reported here: https://towardsdatascience.com/3d-object-classification-in-6-steps-using-kaolin-and-colab-9ecb079143a8

After the first step I receive: "ImportError Traceback (most recent call last) in () 17 import torch 18 from torch.utils.data import DataLoader ---> 19 import kaolin as kal 20 from kaolin import ClassificationEngine 21 from kaolin.datasets import ModelNet

4 frames /usr/local/lib/python3.7/dist-packages/kaolin/ops/batch.py in () 14 15 import torch ---> 16 from . import tile_to_packed_cuda 17 from . import packed_sum_cuda 18

ImportError: /usr/local/lib/python3.7/dist-packages/kaolin/ops/tile_to_packed_cuda.so: undefined symbol: _ZN3c104cuda20getCurrentCUDAStreamEa"

I googled a lot about this but I found nothing useful. Anybody can help me here? Thank you Fabio

[Caenorst]

Hi @feboz please notice that this is pre-reformat. However, you should be able to import kaolin even without GPU, let me explore

[clinsky]

Hi @feboz please notice that this is pre-reformat. However, you should be able to import kaolin even without GPU, let me explore

Having the same issue as described above. Was any fix found for this?

[gkadusumilli]

Hi @feboz That article was written by me. Glad that you find it helpful https://towardsdatascience.com/3d-object-classification-in-6-steps-using-kaolin-and-colab-9ecb079143a8

I have updated the article. There were recent changes in the Kaolin version and the article was written based on the Kaolin release V0.1

You can revisit the article.

[robysurendran]

Still the error persists ModuleNotFoundError Traceback (most recent call last) in () 1 import torch 2 from torch.utils.data import DataLoader ----> 3 import kaolin as kal 4 from kaolin import ClassificationEngine 5 from kaolin.datasets import ModelNet

2 frames /content/kaolin_1/Documents/kaolin-0.1/kaolin/visualize/vis.py in () 20 import kaolin 21 import kaolin as kal ---> 22 import pptk 23 24

ModuleNotFoundError: No module named 'pptk'

---

NOTE: If your import is failing due to a missing package, you can manually install dependencies using either !pip or !apt.

To view examples of installing some common dependencies, click the "Open Examples" button below.

[Caenorst]

Closing for inactivity