Results Error with different obj file

[zaidworks515]

Hello HuayuanSong, Hope You are doing good. I have used your 3d teeth segmentation model and it is performing really well on the obj files I have. I guess I'm missing something while taking the predictions. Following attached are the results and input obj file.
Please guide me through this issue.

Thanks alot,
Have a great day!!!
cc6600a2-f3d8-4971-bcfc-0075a388e570.zip

[HuayuanSong]

Hey there Zaid (: -thanks for checking out the repo!
So, regarding your issue; I suspect that it's due to your example mesh looking a bit different than what the model was trained on. You see, the training samples I used, included more substantial areas besides the teeth themselves, whereas, your example scan does not include the chunky bit.. maybe that's why? Not quite sure. My best advice is to take a look at the original MeshSegNet repo, https://github.com/Tai-Hsien/MeshSegNet, and see if you can train your own model on your scans -and then you can just replace the model file with your model, and the frontend should still work.

Hello HuayuanSong, Hope You are doing good. I have used your 3d teeth segmentation model and it is performing really well on the obj files I have. I guess I'm missing something while taking the predictions. Following attached are the results and input obj file. Please guide me through this issue.

Thanks alot, Have a great day!!! cc6600a2-f3d8-4971-bcfc-0075a388e570.zip

[zaidworks515]

Hi HuayuanSong, hope you're doing great. Actually I need the reference from where I can see how the data is labelled, annotated, segmented then how can we train a model on labelled data, how the model config is. So, I can train a model on my custom dataset

[HuayuanSong]

Hi Zaid,
Thanks for your good wishes. I hope you are doing well, too.

Now, it's been a while.. but here's some of the reference:
https://paperswithcode.com/dataset/teeth3ds
https://github.com/abenhamadou/3DTeethSeg22_challenge
https://3dteethseg.grand-challenge.org/

-and here's a tool that Tai-Hsien used in his original MeshSegNet repo:
https://github.com/Tai-Hsien/Mesh_Labeler

I hope that will help,
/Huayuan

[zaidworks515]

Hi HuayuanSong, hope you're doing well. I have started to gather dataset of interoral scans, and I want to label that data and then train a segmentation model on it. Do you know a suitable tool for it ??
I want the 'labels.json' file like this
{
"labels": [0, 0, 44, 33, 34, 0, 0, 45, 0, .. ,41, 0, 0, 37, 0, 34, 45, 0, 31, 36],
"instances": [0, 0, 10, 2, 12, 0, 0, 9, 0, 0, .. , 10, 0, 0, 8, 0, 0, 9, 0, 1, 8, 13],
}