Prediction on test result is not classifying the body parts

[DavidTu21]

Dear author,

Thank you for your amazing work! I am inferencing/testing on your sample test data (test_scan_002).ply using your pretrained model. However I faces a problem that the model is not correctly predict the body parts.

When I use body part = 4 (and the corresponding pretrained weight best_4_class.pt), I got everything classified as one class:

Similarly, for body part = 6, I got

For body part = 28, I got

I deleted the processed folder each time when changing the body part numbers.

My folder structure looks like:
test_data/
----------raw/
---------------test_scan_002.ply
ply_data/
----------input/
----------------test_scan_002.ply

My environment are: Pytorch 2.2.2, Python 3.8, torch-geometric 2.5.2.

Any insights would be much appreciated! Thank you for your time.

Kind regards,
David

[ayk731]

I got the same output. I'm looking forward to good news.

[ykk8687]

Dear author,

Thank you for your amazing work! I am inferencing/testing on your sample test data (test_scan_002).ply using your pretrained model. However I faces a problem that the model is not correctly predict the body parts.

When I use body part = 4 (and the corresponding pretrained weight best_4_class.pt), I got everything classified as one class:

Similarly, for body part = 6, I got

For body part = 28, I got

I deleted the processed folder each time when changing the body part numbers.

My folder structure looks like: test_data/ ----------raw/ ---------------test_scan_002.ply ply_data/ ----------input/ ----------------test_scan_002.ply

My environment are: Pytorch 2.2.2, Python 3.8, torch-geometric 2.5.2.

Any insights would be much appreciated! Thank you for your time.

Kind regards, David

hi
I found that the pretrained model(best 4_class.pt etc) might have some issues.

I just retrain a new model with default settings and it works fine for me

result: