Xiaoyang Wu
Xiaoyang Wu
Hi, pleaase check whether the processed data is correctly linked to the default data path (Pointcept/data/s3dis)
Hey Lukas, check this testing script for reference! ``` import os import numpy as np import torch import torch.nn.functional as F from collections import OrderedDict from pointcept.datasets import build_dataset from...
Yes, this is for visualization of point cloud embedding. You can modified it we some actually inference propose.
Yeah, you can check my preprocessing code for how to process normal from various dataset. What’s the format of your raw data?
Hi, sorry for the late response, I just recovered from the busy schedule for the CVPR project. PlY can be mesh or point cloud, and some ply already contains normal...
track one of our preprocessing code would be helpful (https://github.com/Pointcept/Pointcept/blob/main/pointcept/datasets/preprocessing/scannet/preprocess_scannet.py)
Hi, normal information can be extracted from raw data like mesh. What kind of resources does your custom dataset contain?
Oh really? I forgot which one, I thought there was no difference.
I think no difference between this two versions?
Hi, I think reading the script we provided for S3DIS and removing unnecessary parts like getting the norm from the raw data of S3DIS is a good choice.