Issue about the size of ModelNet 40

[yangze0930]

Hi， You mentioned that there are 12,311 object instances in the ModelNet 40 dataset in your paper. However, in your provided dataset, I find there are only 3,983 object instances. Could you please provide more details about the dataset. Thank you for your help.

[suhangpro]

This is a very common confusion regarding the ModelNet benchmark. There are 12311 shapes in total, but according to the original authors (in their original CVPR paper: '3D ShapeNets: A Deep Representation for Volumetric Shape Modeling'), the benchmark is using only 100 shapes per category (slightly few for some categories), resulting ~4k shapes in total.

Unfortunately, users on this benchmark very often use the whole dataset for training and evaluation, making direct comparison difficult.

We noted this in the footnote of our experiment section.

[yangze0930]

Thanks for your clarification.