search

FENGGENYU / CVPR2019_PartNet

Code for PartNet: A Recursive Part Decomposition Network for Fine-grained and Hierarchical Shape Segmentation
MIT License
111 stars 18 forks source link

sample selection criteria for partnet-symh dataset #9

Closed lx1037 closed 4 months ago

lx1037 commented 4 months ago

Hi, fenggen

I noticed that there are 4045 samples of airplanes in the ShapeNet dataset, but only 630 samples in the PartNet dataset you released. Did you follow any specific criteria when selecting samples for the PartNet-Symh dataset?

Thank you!

Best regards, ziyu

lx1037 commented 4 months ago

Sorry, I have another question. Thank you for sharing your dataset creation process at [https://github.com/FENGGENYU/CVPR2019_PartNet/issues/5#issuecomment-1176861275], which is very helpful to me. I would also like to ask what method you used to do fine-grained part segmentation in step 1? Thank you very much!

FENGGENYU commented 4 months ago

The data in PartNet dataset are from previous work GRASS, not sampled from Stanford ShapeNet.

Data in PartNet-Symh dataset are from Stanford PartNet dataset.

Step 1 are usually done by mannual segmentation, such as maya and 3ds max.

lx1037 commented 4 months ago

I see. Thanks a lot!