Tao Lu

支持! 感谢作者的辛苦维护

Hi, Xiaobing. The related codes are posted as follows, https://github.com/MCG-NJU/CGA-Net/blob/e7a46cb4379eb3c7979074779909e051f01d9d31/examples/RandLA-Net/main_S3DIS.py#L178-L191

@whuhxb Hi, the result of the code is 68.6% on Area 5 of S3DIS.

Have you changed any experiment settings? If using the default settings, you may try to use some other checkpoint files (I mainly used the last two checkpoint files) for testing.

Sorry for the late reply. I use the last two saved checkpoint files for testing. The improved area is calculated by ``` right_origin = (pred_origin==gt_label) right_cga = (pred_cga==gt_label) improved_area =...

Thanks! The core code is available now.

Hi, thanks for your attention. You may check whether the mpirun is installed correctly according to the instructions. Because there are different versions of mpirun (like open mpi, intel mpi,...

Well, I've met this before and found several reasons may cause this. You can check: 1. whether the process get stuck in loading data (due to wrong data path); 2....

Hi, we add an extra parameter ```r``` to control the devoxelization size, see https://github.com/MCG-NJU/LinK/blob/a91a9a5744062ae5e4f1d689ff3bf125897b4444/segmentation/torchsparse-u/torchsparse/backend/devoxelize/devoxelize_cuda.h#L12 And, you can locate all the associated modifications with a ```diff``` command.

Hi, two ways can generate the depth map: 1. Override the ```color``` with ```distance from xyz to camera_center``` and do rasterization: ``` xyz_dist = torch.norm(xyz - viewpoint_camera.camera_center, dim=-1, keepdims=True) color...