Xiaoyang Wu

Sure, [here](https://connecthkuhk-my.sharepoint.com/:u:/g/personal/wuxy_connect_hku_hk/Ec3MqkjyNkJBiAcvvroqWx4BDZmiSsadeK1wksT0VfWu5Q?e=ZElbVD). The checkpoint is used for codebase released reproduced result.

Sure! The following ShapeNet config for PTv1 was from an older version of our codebase, and you need to migrate it to the released codebase. Also, I did not tune...

Hi, `/config/base/datasets/shapenet_part.py` is just a base config, not a task config. It doesn't contain the necessary components to start a task. You can modify it from a task config located...

Hi, I only provide an `stv1m2` config for `scannet`. You need to rewrite one for s3dis. (You can modify it from `PTv2m2` config)

Readme here (https://github.com/Pointcept/Pointcept?tab=readme-ov-file#offset) will be helpful. For real-world data, as each point cloud contains a different number of points, we have to merge the batch (b) and num_point (n) into...

> My question is, do we actually need RGB value in order to segment? Is colorless point cloud going to have a bad semantic segmentation result? It is not necessary,...

> The input of PTv2 contains 720,000 points, so is this the result of adding up 12 batches? Yes. > Then why is the batch from 0 to 5 (6...

Bohao didn't do that as limited time and resources, but I think it can work well with Waymo. You can modify the config from waymo+spunet

As mix_prob=0.8. we randomly mixup two point cloud.

Oh, thanks for your kind reminder. This is an additional dependence from Point Prompt Training