How to train ToothGroupNetwork

[liufububai]

Hello,

"if you want to train our model, please contact me. The training code will be updated and released during May 2023."

How to train this model and I don't find your email.

Please let me know.

Thanks!

[limhoyeon]

Hello.

I apologize for the delay in making it public, as I have been occupied with graduation preparations. I plan to release it within the next two weeks.

Best regards Lim

On Sat, Jun 3, 2023 at 10:45 AM liufububai @.***> wrote:

Hello,

"if you want to train our model, please contact me. The training code will be updated and released during May 2023."

How to train this model and I don't find your email.

Please let me know.

Thanks!

— Reply to this email directly, view it on GitHub https://github.com/limhoyeon/ToothGroupNetwork/issues/3, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQW6T2STJPKRWVLH2ESB3XDXJKJKBANCNFSM6AAAAAAYY64TGU . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[liufububai]

I meet error on this cd external_libs/pointops && python setup.py install like "File "/.../python3.8/site-packages/torch/utils/cpp_extension.py", line 1555, In_run_ninja_build raise RuntimeError(message) from e RuntimeError: Error compiling objects for extension"

What's your python version?

[limhoyeon]

Maybe this issue is related to cuda compiler(nvcc) and compiler. Are nvcc and compiler(maybe gcc in ubuntu) installed in your environment? My python version is 3.7.

On Sun, Jun 11, 2023 at 6:49 PM liufububai @.***> wrote:

I meet error on this cd external_libs/pointops && python setup.py install like "File "/.../python3.8/site-packages/torch/utils/cpp_extension.py", line 1555, In_run_ninja_build raise RuntimeError(message) from e RuntimeError: Error compiling objects for extension"

What's your python version?

— Reply to this email directly, view it on GitHub https://github.com/limhoyeon/ToothGroupNetwork/issues/3#issuecomment-1586095573, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQW6T2RDOUDHXYPYZZD7T2DXKWIEHANCNFSM6AAAAAAYY64TGU . You are receiving this because you commented.Message ID: @.***>

[liufububai]

I rent a cloud GPU server and CUDA driver is 12.0. Maybe you are right that it is the cuda version and environment setttings problems. I will try to set them as you recommend. Thanks!

[liufububai]

We solve the problem by reinstall CUDA driver. We can generate .json, but we don't quite understand the meaning of label "instances".

We read this link: https://github.com/abenhamadou/3DTeethSeg22_challenge/blob/main/evaluation/export_labeled_3Dscan.py

But, we don't fully understand this code and visualize wrong results. We think the color is assign on vertices. How to recover the color to mesh?

[limhoyeon]

Yes, the challenge data assigns colors to the vertices, and the color of the each face is determined by the vertex colors of each face. Do you have miccai challenge dataset? Maybe you can test the visualization code with miccai challenge dataset.

On Tue, Jun 13, 2023 at 11:46 AM liufububai @.***> wrote:

We solve the problem by reinstall CUDA driver. We can generate .json, but we don't quite understand the meaning of label "instances".

We read this link: https://github.com/abenhamadou/3DTeethSeg22_challenge/blob/main/evaluation/export_labeled_3Dscan.py

But, we don't fully understand this code and visualize wrong results. We think the color is assign on vertices. How to recover the color to mesh? [image: incorrect] https://user-images.githubusercontent.com/11657709/245336896-d4ffebfc-6825-46e4-ab00-fa79514cdda3.jpg

— Reply to this email directly, view it on GitHub https://github.com/limhoyeon/ToothGroupNetwork/issues/3#issuecomment-1588435760, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQW6T2XWIZYPYKO7SELRFZDXK7H6RANCNFSM6AAAAAAYY64TGU . You are receiving this because you commented.Message ID: @.***>

[liufububai]

Sorry, I don't have miccai challenge dataset. So far, I don't figure out how to apply miccai challenge dataset, which looks closing the application of dataset. We find lower jaws look correct, but upper jaws look wrong. Do you think it is caused by jaw data format? And the key of "instances" means the id of teeth for a vertex or the number of teeth in the 3D scan. It is hard to understand the meaning.

[limhoyeon]

Plz check data description! maybe the upper jaw have to be rotated, so that the upper jaw and the lower jaw have same axis.

And the key of "instances" means the id of teeth for a vertex. If the instance number of vertex is "34", this vertex belongs to 34(in fdi tooth numbering system). the order of the "instaces" array is the same as obj file vertex order.

[liufububai]

Thanks. We get results after we rotate upper jaws to the same poses of lower jaws. Another issues is that we find the teeth are not semented completely, see our snapshot: The areas near gingiva are incorrect. Is this issues caused by sampling?

[limhoyeon]

can you give me the sampled point cloud visualization results?

On Fri, Jun 16, 2023 at 2:00 PM liufububai @.***> wrote:

Thanks. We get results after we rotate upper jaws to the same poses of lower jaws. Another issues is that we find the teeth are not semented completely, see our snapshot: [image: ToothGroupNetwork_results1] https://user-images.githubusercontent.com/11657709/246319662-b2285923-f010-418e-b4e0-61522ae6ee56.jpg The areas near gingiva are incorrect. Is this issues caused by sampling?

— Reply to this email directly, view it on GitHub https://github.com/limhoyeon/ToothGroupNetwork/issues/3#issuecomment-1594099959, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQW6T2WOTEGW4W7IMUAOW2TXLPR5PANCNFSM6AAAAAAYY64TGU . You are receiving this because you commented.Message ID: @.***>

[limhoyeon]

plz refer to https://github.com/limhoyeon/ToothGroupNetwork/issues/4

[gxu12]

@liufububai is it possible for you to teach me how to create json file? or can you leave your contact emails? pretty appreciate