Xiaoyang Wu

Hi, after preprocessing, ScanNet, S3DIS, and Structured3D are all organized in the following data structure: ``` dataset |- split1 (e.g., train, val, test in ScanNet or area1, area2, ..., area6...

Another solution is to set `ulimit -n 1048576` in your bashrc. I also encountered a similar issue when I turned my debug platform to a Windows WSL system.

Hi, try to use our another testing strategy as discussed, please see attached here. https://github.com/Pointcept/Pointcept/issues/108#issuecomment-1893709522

Thank you for pointing out this issue!!

`In my example, i refers to the reflection intensity (scalar field, [n, 1] dim) vector. Is it feasible to include the intensity as an additional feature vector in the implementation?`...

Hi, as you align your data with a ScanNet format, I think you can directly modify your config from ScanNet (https://github.com/Pointcept/Pointcept/blob/main/configs/scannet/semseg-pt-v3m1-0-base.py), just align the num_classes, class_name, data_root as your custom...

Hi, it might be because the original point cloud is too huge and there are too much of pieces of sub-point cloud need to be predict during testing (check our...

Hi, refer paper Mix3D and you can set mix_prob = 0 to prevent this.

Hi, the results should have the same density as the original point cloud although we adopt two grid sizes. Pointcept will interpolate the downsampled point cloud to the original density....

> why to have two grid_sample in test configuration. And the mode of first one is "train"? A faster testing strategy which mixes interpolation (original -> 0.025) and our precise...